The Circumstellar Environments of Young Stars at AU Scales

Rafael Millan-GabetCalifornia Institute of Technology

Fabien MalbetLaboratoire d’Astrophysique de Grenoble

Rachel AkesonCalifornia Institute of Technology

Christoph LeinertMax-Planck-Institut fur Astronomie

John MonnierUniversity of Michigan

Rens WatersUniversity of Amsterdam

We review recent advances in our understanding of the inner-most regions of the cir-cumstellar environment around young stars made possible by the technique of long baselineinterferometry at infrared wavelengths. Near-infrared observations directly probe the locationof the hottest dust. The characteristic sizes found are much larger than previously thought, andstrongly correlate with the luminosity of the central young stars. This relationhas motivatedin part a new class of models of the inner disk structure. The first mid-infrared observationshave probed disk emission over a larger range of scales, and spectrally resolved interferometryhas for the first time revealed mineralogy gradients in the disk. These newmeasurementsprovide crucial information on the structure and physical properties ofyoung circumstellardisks, as initial conditions for planet formation. In addition to summarizing these pioneeringobservations, we expose the many open questions that accompany the impressive progressmade, and anticipate the experimental and modelling efforts that promise tohelp elucidate thediverse phenomena associated with the close circumstellar environmentof young stars.

1. INTRODUCTION

Stars form from collapsing clouds of gas and dust andin their earliest infancies are surrounded by complex en-vironments that obscure our view at optical wavelengths.As evolution proceeds, a stage is revealed with three maincomponents: the young star, a circumstellar disk and an in-falling envelope. Eventually, the envelope dissipates, andthe emission is dominated by the young star-disk system.Later on, the disk also dissipates to very tenuous levels.It is out of the young circumstellar disks that planets areexpected to form, and therefore understanding their phys-ical conditions is necessary before we can understand theformation process. Of particular interest are the inner fewAU (Astronomical Unit), corresponding to formation sitesof Terrestrial type planets, and to migration sites for gasgiants presumably formed further out in the disk (see thechapter by Udry, Fisher and Queloz).

A great deal of direct observational support exists for the

scenario outlined above, and in particular for the existenceof circumstellar disks around young stars (see the chaptersby Dutrey, Guilloteau & Ho and by Menard et al.). In addi-tion, the spectroscopic and spectro-photometric character-istics of these systems (i.e. spectral energy distributions,emission lines) are also well described by the disk hypoth-esis. However, state of the art optical and millimeter-waveimaging typically probe scales of 100−1000s of AU withresolutions of 10s of AU, and models of unresolved obser-vations are degenerate with respect to the spatial distribu-tion of material. As a result, our understanding of even themost general properties of the circumstellar environment atfew AU or smaller spatial scales is in its infancy.

Currently, the only way to achieve sufficient angularresolution to directly reveal emission within the inner AUis through optical interferometry at visible and infraredwavelengths. An interferometer with a baseline length ofB = 100 m (typical of current facilities) operating atnear to mid-infrared wavelengths (typically H to N bands,

1

λ0 = 1.65 − 10µm) probes1800 − 300 K material andachieves an angular resolution∼ λ0/2B ∼ 3 − 26 mil-liarcseconds (mas), or 0.5−4 AU at a distance typical ofthe nearest well known star forming regions (150 pc). Thisobservational discovery space is illustrated in Fig. 1, alongwith the domains corresponding to various young stellar ob-ject (YSO) phenomena and complementary techniques andinstruments. Optical interferometers are ideally suited todirectly probe the innermost regions of the circumstellar en-vironment around young stars, and indeed using this tech-nique surprising and rapid progress has been made, as theresults reviewed in this chapter will show.

Optical, as well as radio, interferometers operate by co-herently combining the electromagnetic waves collectedby two or more telescopes. Under conditions that ap-ply to most astrophysical observations, the amplitude andphase of the resulting interference patterns are relatedvia a two-dimensional Fourier transform to the bright-ness distribution of the object being observed. For a de-tailed description of the fundamental principles, variationsin their practical implementation, and science highlightsin a variety of astrophysics areas we refer the reader tothe reviews byMonnier (2003) andQuirrenbach(2001).The reader may also be interested in consulting the pro-ceedings of topical summer schools and workshops suchas Lawson(2000), Perrin and Malbet(2003), Garcia etal. (2003) andParesce et al.(2006) (online proceedings ofthe Michelson Summer Workshops may also be found athttp://msc.caltech.edu/michelson/).

While interferometers are capable of model-independentimaging by combining the light from many telescopes, mostresults to date have been obtained using two-telescopes(single-baseline). The main characteristics of current fa-cilities involved in these studies are summarized in Table 1.Interferometer data, even with sparse spatial frequency cov-erage, provides direct constraints about source geometryand thus contains some of the power of direct imaging.However, with such data alone only the simplest single-component objects can be constrained. Therefore, moretypically, a small number of fringe visibility data points arecombined with a spectral energy distribution (SED) for fit-ting simple physically-motivated geometrical models, suchas a Gaussian profile or ring emission (depending on thecontext). This allows us to determine “characteristic sizes”at the wavelength of observation for many types of YSOs(e.g. T Tauri, Herbig Ae/Be, FU Orioni). In addition, thevisibility amplitudes in a given spectral band can be com-pared to predictions of specific physical models. Realisticmodels models are usually defined by many more parame-ters than can be uniquely constrained by the interferometerdata alone, however in combination with simultaneous SEDfitting, this exercise has allowed to rule out certain classesof models and provided useful constraints for models thatcan be made to reproduce the spectral and spatial observ-ables.

A note on nomenclature: Before the innermost disk re-gions could be spatially resolved, as described in this re-

view, disk models were tailored to fit primarily the unre-solved spectro-photometry. The models used can be gener-ally characterized as being geometrically thin and opticallythick, with simple radial temperature power laws with expo-nentsq = −0.75 (flat disk heated by accretion and/or stellarirradiation,Lynden-Bell and Pringle, 1974) or flatter (flareddisks, Kenyon and Hartmann, 1987; Chiang and Goldre-ich, 1997). In this review, we will refer to these classes ofmodels as “classical”, to be distinguished from models witha modified inner disk structure that have recently emergedin part to explain the NIR interferometer measurements.

The outline of this review is as follows. Sections 2 and 3summarize the observational state of the art and emerginginterpretations, drawing a physically motivated distinctionbetween inner and outer disk regions that also naturally ad-dresses distinct wavelength regimes and experimental tech-niques. We note that these developments are indeed recent,with the first preliminary analyses of interferometric obser-vations of YSOs being presented (as posters) at the Proto-stars and Planets IV Conference in 1998. In Section 4 wehighlight additional YSO phenomena that current facilitiesare addressing, such as the disk-wind connection and youngstar mass determination, using combination of interferom-etry with high resolution line spectroscopy. In Section 5we present a top-level summary of the main observationalresults, and highlight remaining important open questionslikely to be experimentally and theoretically addressed incoming years. In Section 6 we describe additional phenom-ena yet to be explored and new instrumental capabilities thatpromise to enable this future progress.

mm interf.

8/10m ground−based

telescopes

Inte

rfero

metry

Optic

al

ALMA

HST

1500

150

15

1.5

0.015

Du

st

su

bli

ma

tio

n

Thermal disk emission

0.1 mas

0.151 mas

0.01

10

1

0.1

Disk

envelopes

mic

ro−

jets

Ma

gn

eto

−

sp

he

re

Wavelength (microns)

Stars

op

tic

al

jets

Cold dust

Gas reservoir

Win

ds

Inn

er

ga

s d

isk

An

gu

lar

reso

luti

on

(arc

sec)

Lin

ear s

cale

(AU

, for d

= 1

50 p

c)

from disks

Scattering

1 100010010

Fig. 1.—Observational phase-space (spectral domain and angu-lar resolution) for optical interferometers, and for complementarytechniques (shaded polygons). Also shown are the main physicalphenomena associated with young stellar objects, in rectangularboxes approximately located over the most relevant phase-spaceregions. On the right vertical axis, angular resolution is translatedto linear scales assuming a distance of 150 pc.

2. THE INNER DISK

2

TABLE 1

LONG BASELINE OPTICAL INTERFEROMETERSINVOLVED IN YSO RESEARCH

Facility Instrument Wavelength Number of Telescope Baseline Range Best Resolutionc

Coveragea Telescopesb Diameter (m) (m) (mas)

PTI V 2 1.6 − 2.2µm [44] 2 [3] 0.4 80− 110 1.5IOTA V 2, IONIC3 1.6 − 2.2µm 3 0.4 5− 38 4.5ISI Heterodyne 11µm 3 1.65 4− 70 16.2KI V 2 1.6 − 2.2µm [22] 2 10 85 2.0KI Nuller 8 − 13µm [34] 2 10 85 9.7VLTI MIDI 8 − 13µm [250] 2 [8] 8.2 / 1.8 8− 200 4.1VLTI AMBER 1 − 2.5µm [104] 3 [8] 8.2 / 1.8 8− 200 0.6CHARA V 2 1.6 − 2.2µm 2 [6] 1 50− 350 0.4

NOTE.—V 2 refers to a mode in which only the visibility amplitude is measured, often implemented in practice as ameasurement of the (un-biased estimator) square of the visibility amplitude.

aThe maximum spectral resolution available is given in parenthesis

bThe number of telescopes that can be simultaneously combined is given,along with the total number of telescopesavailable in the array in parenthesis.

cIn each case, the best resolution is given as half the fringe spacing,λmin/(2Bmax), for the longest physical baselinelength and shortest wavelength available.

References. — [1] PTI,Palomar Testbed Interferometer, Colavita et al1999. [2] IOTA, Infrared-Optical TelescopeArray, Traub et al.2004. [3] ISI,Infrared Spatial Interferometer, Hale et al.2000. [4] KI,Keck Interferometer, Colavita,Wizinovich and Akeson2004. [5] KI Nuller,Serabyn et al.2004. [6] MIDI, MID-infrared Interferometric instrument,Leinert2004. [7] AMBER,Astronomical Multiple BEam Recombiner, Malbet et al.2004. [8] CHARA,Center for HighAngular Resolution Astronomy interferometer, ten Brummelaar et al.2005.

3

At near-infrared (NIR,∼ 1− 2.4µm) wavelengths opti-cal interferometers probe the hottest circumstellar materiallocated at stellocentric distances. 1 AU, comparable to thelocation of terrestrial planets in more evolved systems. Inthis review these regions are referred to as theinner disk.For technical reasons, it is at NIR wavelengths that inter-ferometric measurements of YSOs first became possible,and provided the first direct probes of inner disk proper-ties. Until only a few years ago, relatively simple modelsof accretion disks were adequate to reproduce most observ-ables. However, the new observations with higher spatialresolution have revealed a richer set of phenomena.

2.1 Herbig Ae/Be Objects

In Herbig Ae/Be (HAeBe) objects, circumstellar mate-rial (proto-planetary disk, remnant envelope, or both) sur-rounds a young star of intermediate mass (∼ 2 − 10M�,see e.g. the review byNatta et al.2000). A significantnumber of these objects fall within the sensitivity limits ofsmall interferometers, hence the first YSO studies contain-ing relatively large samples focused on objects in this class.

AB Aur was the first “normal” (i.e. non-outburst, seeSection 2.3) YSO to be resolved at NIR wavelengths. In ob-servations at theInfrared Optical Telescope Array(IOTA)AB Aur appeared clearly resolved in the H and K spec-tral bands on 38 m baselines, by an amount correspondingto a characteristic diameter of 0.6 AU for the circumstel-lar NIR emission (Millan-Gabet et al., 1999). The mea-sured size was unexpectedly large in the context of then-current disk models of HAeBe objects (e.g.Hillenbrand etal., 1992) which predicted NIR diameter of 0.2 AU based onoptically-thick, geometrically-thin circumstellar diskwitha small dust-free inner hole. This conclusion was rein-forced by the completed IOTA survey ofMillan-Gabet etal. (2001), finding characteristic NIR sizes of 0.6−6 AUfor objects with a range of stellar properties (spectral typesA2–09).

Due to the limited position angle coverage of thesefirst single-baseline observations, the precise geometry ofthe NIR emission remained ambiguous.Millan-Gabet etal. (2001) found that the NIR sizes were similar at 1.65 and2.2µm, suggesting a steep temperature gradient at the in-ner disk edge. These early observations revealed no directevidence for “disk-like” morphologies; indeed, for the fewobjects with size measurements at multiple position angles,the data indicated circular symmetry (as if from sphericalhalos or face-on disks, but not inclined disks).

The first unambiguous indication that the NIR emissionarises in disk-like structures came from single-telescopeimaging of the highly-luminous YSOs LkHα 101 andMWC 349-A, using a single-telescope interferometrictechnique (aperture masking at the Keck I telescope).MWC 349-A appeared clearly elongated (Danchi et al., 2001)and LkHα 101 presented an asymmetric ring-like morphol-ogy, interpreted as a bright inner disk edge occulted onone side by foreground cooler material in the outer re-

gions of the flared disk (Tuthill et al., 2001). Moreover,the location of the LkHα 101 ring was also found to beinconsistent with predictions of classical disk models andthese authors suggested that an optically-thin inner cavity(instead of optically-thick disk midplane) would push thedust sublimation to larger radii consistent with the new sizemeasurements.

Could simple dust sublimation of directly heated dustbe setting the NIR sizes of other HAeBe objects as well?The idea was put to the test byMonnier and Millan-Gabet(2002). Inspired by the LkHα 101 morphology andinterpretation, these authors fit a simple model consistingofa central star surrounded by a thin ring for all objects thatwere measured interferometrically at the time (IOTA andAperture Masking as referenced above, plus first YSO re-sults from thePalomar Testbed Interferometer, PTI,Akesonet al., 2000). Indeed, the fitted ring radii are clearly corre-lated with the luminosity of the central star, and follow theexpected relationR ∝ L

1/2

? . Further, by using realistic dustproperties it was also found that the measured NIR sizesare consistent with the dust sublimation radii of relativelylarge grains (& 1µm) with sublimation temperatures in therange 1000 K−2000 K.

Following Monnier and Millan-Gabet(2002), in Fig. 2we have constructed an updated diagram of NIR sizevs.central luminosity based on all existing data in the literature.We include data for both HAeBe and T Tauri objects, thelatter being discussed in detail in the next Section; and wealso illustrate schematically the essential ingredients (i.e.location of the inner dust disk edge) of the models to whichthe data are being compared. For HAeBe objects, in addi-tion to the references above, we include additional measure-ments from PTI (Eisner et al., 2003, 2004) and from recentobservations using theKeck Interferometer(KI) (Monnieret al., 2005). For HAeBe objects, disk irradiation is dom-inated by the stellar luminosity, and therefore we neglectheating by accretion shock luminosity (which may play asignificant role for the most active lower stellar luminosityT Tauri objects, as will be seen in the next section).

Indeed, it can be seen that for HAe and late HBe ob-jects, the measured NIR sizes are tightly contained, overmore than two decades in stellar luminosity, within the sub-limation radii of directly heated grey dust of sublimationtemperature in the range 1000 K−1500 K under the as-sumption that dust grains radiate over the full solid angle(e.g., no back-warming). If instead we assume that the dustgrains emit only over 2π steradian (e.g., full back-warming)then the corresponding sublimation temperature range is1500 K−2000 K.

The most luminous objects (the early spectral type HBeobjects) on the other hand have NIR sizes in good agree-ment with the classical model, indicating that for these ob-jects the disk extends closer in to the central star.Mon-nier and Millan-Gabet(2002) hypothesized that even op-tically thin low-density gas inside the dust destruction ra-dius can scatter UV stellar photons partly shielding the dust

4

at the inner disk edge and reducing the sublimation radius.Eisner et al.(2003, 2004) argue instead that the distinc-tion may mark a transition from magnetospheric accretion(late types) to a regime for the early types where optically-thick disks in fact do extend to the stellar surface, eitheras a result of higher accretion rates or weak stellar mag-netic fields. We note that using polarimetry across Hα

lines originating in the inner gas disk at∼ few R?, Vink etal. (2005) (and references therein) also find (in addition toflattened structures consistent with disk-like morphology) amarked difference in properties between HBe objects andHAe and T Tauri objects; consistent with the presence ofan un-disrupted gas disk on small scales for HBe objects,while the less massive objects (HAe and T Tauri) show ev-idence for line photons scattering off an external rotatingdisk. However, notall high luminosity HBe objects are inbetter agreement with the classical model; there are notableexceptions (e.g. LkHα 101, MWC 349-A, see Fig. 2) andMonnier et al.(2005) have noted that these may be moreevolved systems, where other physical processes, such asdispersal by stellar winds or erosion by photo-evaporation,have a dominant effect in shaping the inner disk.

In parallel with these developments, and in the contextof detailed modelling of the NIR SED,Natta et al.(2001)similarly formulated the hypothesis that the inner regionsofcircumstellar disks may be largely optically thin to the stel-lar photons such that a directly illuminated “wall” forms atthe inner dust disk edge. The relatively large surface areaof this inner dust wall is able to produce the required levelsof NIR flux (the so-called NIR SED “bump”). The sim-ple blackbody surface of the initial modelling implementa-tion has since been replaced by more realistic models whichtake additional physical effects into account: self-consistenttreatment of radiative transfer and hydrostatic vertical struc-ture of the inner wall and corresponding outer disk shad-owing (Dullemond et al., 2001), optical depth and scaleheight of gas inside the dust destruction radius (Muzerolle etal., 2004), and inner rim curvature due to density-dependentdust sublimation temperature (Isella and Natta, 2005).

Independent of model details however, it is clear that theoptically-thin cavity disk model with a “puffed-up” innerdust wall can explain a variety of observables: the ring-likemorphology, the characteristic NIR sizes and the NIR SED.We also emphasize that although the discussion of the in-terferometer data presented above was centered around thesize-luminosity diagram, the same conclusions have beenobtained by modelling SEDs and visibility measurementsfor individual sources using physical models of both classi-cal and puffed-up inner rim models.

Although the characteristic NIR sizes have now beenwell established for a relatively large sample, few con-straints still exist as to the actual geometry of the resolvedNIR emission. In this respect, most notable are the PTIobservations ofEisner et al.(2003, 2004) which providedthe first evidence for elongated emission, consistent withinclined disks. Further, the inclinations found can explainthe residual scatter (50%) in the size-luminosity relationfor

HAe and late HBe systems (Monnier et al., 2005).It is important to note that three different interferome-

ters have now reached the same conclusions regarding thecharacteristic NIR sizes of HAeBe objects, in spite of vastlydifferent field-of-views (FOV) for the three instruments (3,1, and 0.05arcsecFWHM for IOTA, PTI and KI respec-tively). This indicates that, although often found aroundthese objects (e.g.Leinert et al., 2001), large scale (scat-tered) emission entering the FOV – which would lead to in-correct large characteristic size estimates if not explicitelyaccounted for in a composite model – does not dominate themeasurements; a possibility that plagued the early interpre-tations, and still needs to be considered as a possible (small)effect for some systems.

2.2 T Tauri Objects

Only the brightest T Tauri objects can be observed withthe small aperture interferometers, so early observationswere limited in number. The first observations were takenat the PTI and were of the luminous sources T Tau N andSU Aur (Akeson et al., 2000, 2002). Later observations(Akeson et al., 2005) added DR Tau and RY Tau to the sam-ple, and employed all three PTI baselines to constrain thedisk inclination. Using simple geometrical models (or sim-ple power-law disk models) the characteristic NIR sizes (ordisk inner radii) were found to be larger than predicted bySED-based models (e.g. Malbet and Bertout, 1995), simi-lar to the result for the HAe objects. For example, inclinedring models for SU Aur and RY Tau yield radii of 10 R?

and 11 R? respectively, substantially larger than magnetictruncation radii of 3–5 R? expected from magnetosphericaccretion models (Shu et al., 1994). If the inner dust diskdoes not terminate at the magnetic radius, what then setsits radial location? Interestingly, these initial T Tauri ob-servations revealed NIR sizes that were also consistent withsublimation radii of dust directly heated by the central star,suggesting perhaps that similar physical processes, decou-pled from magnetospheric accretion, set the NIR sizes ofboth T Tauri and HAe objects.Lachaume et al.(2003) werethe first to use physically realistic models to fit the new visi-bility data, as well as the SEDs. Using a two-layer accretiondisk model, these authors find satisfactory fits for SU Aur(and FU Ori, see Section 2.3), in solutions that are charac-terized by the mid-plane temperature being dominated byaccretion, while the emerging flux is dominated by repro-cessed stellar photons.

Analysis of T Tauri systems of more typical luminosi-ties became possible with the advent of large aperture in-frared interferometers. Observations at the KI (Colavita etal., 2003;Eisner et al., 2005;Akeson et al., 2005) continuedto find large NIR sizes for lower luminosity stars, in manycases even larger than would be expected from extrapola-tion of the HAe relation.

Current T Tauri measurements and comparison with dustsublimation radii are also summarized in Fig. 2.Eisneret al. (2005), similar to their analysis of HAeBe objects,

5

found better agreement with puffed-up inner wall models,and in Fig. 2 we use their derived inner disk radii. Follow-ing Muzerolle et al.(2003) andD’Alessio et al.(2004), thecentral luminosity for T Tauri objects includes the accre-tion luminosity, released when the magnetic accretion flowencounters the stellar surface, which can contribute to thelocation of the dust sublimation radius for objects experi-encing the highest accretion rates (for the sample shown,accretion luminosity only affects the location of three ob-jects significantly: AS205–A, PX Vul and DR Tau, havingLaccretion/L? ∼ 10.0, 1.8, 1.2 respectively.)

It can be seen that many T Tauri objects follow theHAe relation, although several are even larger given thecentral luminosity (the T Tauri objects located above the1000 K solid line in Fig. 2 are: BP Tau, GM Aur, LkCa 15,RW Aur).

It must be noted however that current measurements ofthe lowest luminosity T Tauri objects have relatively largeerrors. Further, the ring radii plotted in Fig. 2 depend on therelative star/disk NIR flux contributions, most often derivedfrom an SED decomposition which for T Tauri objects isconsiderably more uncertain than for the HAeBe objects,due to their higher photometric variability, and the fact thatthe stellar SEDs peak not far (∼ 1µm) from the NIR wave-lengths of observation (as an exemption, an improved ap-proach was taken byEisner et al.(2005), who derived stel-lar fluxes at the KI wavelength of observation from opti-cal veiling measurements and extrapolation to NIR wave-lengths based on model atmospheres). These caveats il-lustrate the importance of obtaining more direct estimatesof the excess NIR fluxes, via veiling measurements (e.g.Johns-Krull and Valenti, 2001 and references therein).

Under the interpretation that the T Tauri NIR sizes tracethe dust sublimation radii, different properties for the dustlocated at the disk inner edge appear to be required to ex-plain all objects. In particular, if the inner dust disk edgebecomes very abruptly optically thick (e.g., strong back-warming), the sublimation radii become larger by a factorof ×2, in better agreement with some of the low luminosityT Tauri objects (dashed lines in Fig. 2). As noted byMon-nier and Millan-Gabet(2002), smaller grains also lead tolarger sublimation radii, by a factor

√QR, whereQR is the

ratio of dust absorption efficiencies at the temperatures ofthe incident and re-emitted field (QR ' 1 − 10 for grainsizes1.0 − 0.03µm, the incident field of a stellar temper-ature of 5000 K typical of T Tauri stars, and re-emission at1500 K).

Alternatively, an evolutionary effect could be contribut-ing to the size-luminosity scatter for the T Tauri sample.Akeson et al.(2005) note that the size discrepancy (mea-suredvs. sublimation radii) is greatest for objects withthe lowest ratio of accretion to stellar luminosity. If ob-jects with lower accretion rates are older (e.g.Hartmann etal., 1998) then other physical processes such as disk disper-sal (Clarke et al., 2001, see also the review byHollenbach etal., 2000) may be at play for the most “discrepant” systemsmentioned above (the size derived for GM Aur contains ad-

ditional uncertainties, given the evidence in this system fora dust gap at several AU,Rice et al., 2003, and the possibil-ity that it contributes significant scattered NIR light).

In detailed modelling work of their PTI data,Akesonet al. (2005) used Monte Carlo radiative transfer calcu-lations (Bjorkman and Wood, 2001) to model the SEDsand infrared visibilities of SU Aur, RY Tau and DR Tau.These models addressed two important questions concern-ing the validity of the size interpretation based on fits tosimple star+ ring models: (1) Is there significant NIR ther-mal emission from gas inside the dust destruction radius?and, (2) is there significant flux arising in large scale (scat-tered) emission? The modelling code includes accretionand shock/boundary luminosity as well as multiple scatter-ing and this technique naturally accounts for the radiativetransfer effects and the heating and hydrostatic structureofthe inner wall of the disk. In these models, the gas diskextends to within a few stellar radii and the dust disk ra-dius is set at the dust sublimation radius. For SU Aur andRY Tau, the gas within the inner dust disk was found tocontribute substantially to the NIR emission. Modellingof HAeBe circumstellar disks byMuzerolle et al.(2004)found that emission from the inner gas exceeded the stel-lar emission for accretion rates> 10−7 M� yr−1. Thus,even as a simple geometrical representation, a ring modelat the dust sublimation temperature may be too simplisticfor sources where the accretion luminosity is a substantialfraction of the total system luminosity. Moreover, even ifthe NIR gas emission were located close to the central starand were essentially unresolved, its presence would affectthe SED decomposition (amount of NIR excess attributed tothe dust “ring”), leading to incorrect (under-estimated) ringradii. The second finding from the radiative models is thatthe extended emission (here meaning emission from scaleslarger than 10 milliarcseconds) was less than 6% of the totalemission for all sources, implying that scattered light is nota dominant component for these systems.

Clearly, the new spatially resolved observations just de-scribed, although basic, constitute our only direct view intothe physical conditions in young disks on scales. 1 AU,and therefore deliver crucial ingredients as pre-conditionsfor planet formation. As pointed out byEisner et al.(2005)the interpretation that the measured NIR sizes correspondto the location of the innermost disk dust implies that dustis in fact present at Terrestrial planet locations. Conversely,the measured radii imply that no dust exists, and thereforeplanet formation is unlikely, inside these relatively large∼0.1–1 AU inner dust cavities. Finally, that the dust diskstops at these radii may also have implications for planetmigration, depending on whether migration is halted at theinner radius of the gas (Lin et al., 1996) or dust (Kuchnerand Lecar, 2002) disk (see also the chapter by Paploizou etal.).

2.3 Detailed Tests of the Disk Hypothesis: FU Orionis

FU Orionis objects are a rare type of YSO believed to

6

100 102 104 106

Lstar + Laccretion (Lsolar)

0.01

0.10

1.00

10.00

100.00

Rin

g R

adiu

s (A

U)

HAe ObjectsHBe ObjectsTTS Objects

Sublimation radius: direct dust heating, backwarmingSublimation radius: direct dust heating, no backwarmingSublimation radius: standard model

1000K

1500K1000K

1500K

1000K

1500Koptically−thin gas

(puffed−up

optically−thickgas

gas + dust

gas + dust

"Standard" Disk Model − oblique disk heating

Dust SublimationRadius Rs

(magnetosphericaccretion?)

(magnetosphericaccretion?)

inner wall?)

flared disk surfaceDust Sublimation

flared disk surface

Star

Star

Radius Rs

Direct heating of inner dust disk

Fig. 2.— Measured sizes of HAeBe and T Tauri objectsvs. central luminosity (stellar+ accretion shock); and comparison withsublimation radii for dust directly heated by the central luminosity (dashedlines) and for the oblique heating implied by the classicalmodels (dotted line). A schematic representation of the key features of inner disk structure in these two classes of models is also shown.Data for HAeBe objects are from: IOTA (Millan-Gabet et al., 2001), Keck aperture masking (Danchi et al., 2001,Tuthill et al., 2001),PTI (Eisner et al., 2004) and KI (Monnier et al., 2005). Data for T Tauri objects are from PTI (Akeson et al., 2005) and KI (Akeson etal., 2005b,Eisner et al., 2005). For clarity, for objects observed at more than one facility, we include only the most recent measurement.

be T Tauri stars surrounded by a disk that has recently un-dergone an episode of accretion rate outburst (see e.g. thereview byHartmann et al., 1996). During the outburst, thesystem brightens by several visual magnitudes, followedby decade-long fading. The disk luminosity dominates theemission at all wavelengths, and is expected to be well rep-resented by thermal emission by disk annulii that follow thecanonical temperature profileT ∝ r−3/4. With this pre-scription, the model has few parameters that are well con-strained by the SEDs alone (except for inclination effects).In principle then, these systems are ideal laboratories fortesting the validity of disk models.

The very first YSO to be observed with an optical inter-ferometer was in fact the prototype for this class, FU Ori-onis itself (Malbet et al., 1998). Indeed, the first K-bandvisibility amplitudes measured by the PTI agreed well withthe predictions of the canonical disk model. This basic con-clusion has been re-inforced in a more recent study usingmulti-baseline, multi-interferometer observations (Malbetet al., 2005). The detailed work on FU Ori will ultimatelyclose the debate on its nature: current evidence favors thatit is a T Tauri star surrounded by a disk undergoing mas-sive accretion (Hartmann and Kenyon, 1985, 1996) ratherthan a rotating supergiant with extreme chromosphere ac-tivity (Herbig et al., 2003); and under that interpretationthe physical (temperature law) and geometrical (inclinationand orientation) parameters describing the disk have beenestablished with unprecedented detail.

The total number of known objects in this class is small(e.g. 20 objects in the recent compilation byAbrahamet al., 2004), and the subset that is observable by current

optical interferometers is even smaller (about 6 objects,typically limited by instrumental sensitivity at short wave-lengths0.55 − 1.25µm).

Three more FU Orionis objects have been resolved atK-band: V1057 Cyg (PTI,Wilkin and Akeson, 2003; KIMillan-Gabet et al., 2006), V1515 Cyg and ZCMa-SE(Millan-Gabet et al., 2006). In contrast to the FU Ori case,these three additional objects appear more resolved in theNIR than expected from aT ∝ r−3/4 model, and sim-ple exponent adjustments of such a single power-law modeldoes not allow simultaneous fitting of the visibilities andSEDs. On the other hand, these objects are also knownto possess large mid-infrared flux in excesses of emissionby flat disks, usually attributed to a large degree of outerdisk flaring or, more likely, the presence of a dense dustenvelope (Kenyon and Hartmann, 1991). The low visibil-ities, particularly for V1057 Cyg and V1515 Cyg, can beexplained by K-band flux, in addition to the thermal diskemission, resulting from scattering through envelope ma-terial over scales corresponding to the interferometer FOV(50 mas FWHM). Contrary to initial expectations then, thelikely complexity of the circumstellar environment of mostFU Orionis objects will require observations at multiplebaselines and wavelengths in order to disentangle the pu-tative multiple components, discriminate between the com-peting models, and perform detailed tests of accretion diskstructure.

3. THE OUTER DISK

At mid-infrared (MIR) wavelengths, and for 50−100 mbaselines, optical interferometers probe the spatial distribu-

7

tion and composition of∼ few 100 K circumstellar gas anddust, with15 − 50 mas resolution, or2 − 8 AU at a dis-tance of 150 pc. The MIR radiation of circumstellar disks inthe N-band (8 − 13µm) comes from a comparatively widerange of distances from the star, see Fig. 3, and we nowfocus the discussion on circumstellar dust in this several-AU transition region between the hottest dust in the subli-mation region close to the star (Section 2), and the muchcooler disk regions probed by mm interferometry (see thechapter by Dutrey et al.) and by scattered light imaging(see the chapter by Menard et al.). As a matter of interpre-tational convenience, at these longer MIR wavelengths theflux contribution from scattered light originating from theserelatively small spatial scales is negligible, allowing tocon-sider fewer model components and physical processes.

3.1 Disk Sizes and Structure

The geometry of proto-planetary disks is of great im-portance for a better understanding of the processes of diskevolution, dissipation and planet formation. The compo-sition of the dust in the upper disk layers plays a crucialrole in this respect, since the optical and UV photons ofthe central star, responsible for determining the local disktemperature and thus the disk vertical scale-height, are ab-sorbed by these upper disk layer dust particles. The diskscale-height then plays a pivotal role in the process of dustaggregation and settling, and therefore impacts the globaldisk evolution. In addition, the formation of larger bodiesin the disk can lead to structure, such as density waves andgaps. Clearly, high spatial resolution MIR imaging will beessential in order to probe these signs of planet formation.

Spatially resolved observations of disks in the MIR us-ing single telescopes have been limited to probing relativelylarge scale (100s AU) thermal emission in evolved disksaround main-sequence stars (e.g.Koerner et al., 1998;Jayawardhana, 1998) or scattering by younger systems(e.g. McCabe et al., 2003). Single-aperture interferomet-ric techniques (e.g. Nulling,Hinz et al., 2001; Liu etal., 2003, 2005 or Segment-Tilting,Monnier et al., 2004b)as well as observations at theInfrared Spatial Interferome-ter (ISI, Tuthill et al., 2002) have succeeded in resolving thebrightest young systems on smaller scales for the first time,establishing MIR sizes of∼ 10s AU. Although the observedsamples are still small, and thus far concern mostly HAeobjects, these initial measurements have already provideda number of interesting, albeit puzzling results: MIR char-acteristic sizes may not always be well predicted by simple(flared) disk models (some objects are “under-sized” com-pared to these predictions, opposite to the NIR result), andmay not always connect in a straightforward way with mea-surements at shorter (NIR) and longer (mm) wavelengths.

These pioneering efforts were however limited in res-olution (for the single-aperture techniques) or sensitivity(for the heterodyne ISI), and probing the MIR emissionon small scales has only recently become possible withthe advent of the new-generation long baseline MIR instru-

ments (VLTI/MIDI, Leinert, 2004; KI/Nuller, Serabyn etal., 2004).

Two main factors determine the interferometric signa-ture of a dusty protoplanetary disk in the MIR: The overallgeometry of the disk, and the composition of the dust inthe disk “atmosphere.” The most powerful way to disten-tangle these two effects is via spectrally resolved interfer-ometry, and this is what the new generation of instrumentssuch as MIDI at the VLTI are capable of providing. Clearly,some information can be obtained from spatially unresolvedobservations – the overall SED constrains the distributionof material with temperature (or radius), and MIR spec-troscopy provides the properties of dust. However, strongradial gradients in the nature of the dust are expected, bothin terms of size and chemistry, that can only be observedwith spatially resolved observations.

First results establishing the MIR sizes of young circum-stellar disks were obtained by VLTI/MIDI for a sample ofseven HAeBe objects (Leinert et al.2004). The charac-teristic sizes measured, based on visibilities at the wave-length of12.5µm, are in the range 1–10 AU. Moreover, asshown in Fig. 4, they are found to be well correlated withthe IRAS [12]–[25] color, with redder objects having largerMIR sizes. These observations lend additional support tothe SED-based classification ofMeeus et al.(2001) andDullemond and Dominik(2004), whereby redder sourcescorrespond to disks with a larger degree of flaring, whichtherefore also emit thermally in the MIR from larger radii,compared to sources with flatter disks.

A more detailed and powerful analysis is possible byexploiting the spectral resolution capabilities of the MIDIinstrument. Based on the reddest object in their sample(HD 100546), the spectrally resolved visibilities also sup-port the theMeeus et al.(2001) classification: this ob-ject displays a markedly steeper visibility drop between8 − 9µm, and lower visibilities at longer wavelengths thanthe rest of the objects in the sample, implying the expectedlarger radius for the MIR emission. These first observationshave also been used to test detailed disk models that wereoriginally synthesized to fit the SEDs of individual objects(Dominik et al., 2003); without any feedback to the modelsLeinert et al.(2004) find encouraging qualitative agreementbetween the predicted spectral visibility shapes and the in-terferometer data.

It remains to be seen however whether detailed fine-tuning of all available information will solve the remain-ing discrepancies. Thus far, a parametrised disk modelsatisfying both the spectral and interferometric constraintshas been presented only for the somewhat unusual object51 Oph (Gil at al., 2005). However, the solution foundrequires a high accretion rate inconsistent with other evi-dence, signaling the need for further physical characteriza-tion. In general, changes in both disk geometry and dustcomposition strongly affect the MIR spectral visibilities.Howevervan Boekel et al.(2005) have shown that the prob-lem is tractable and even just a few well-chosen baselines(and wavelengths) permit to test some of the key features of

8

recently proposed disk models, such as the presence of theputative puffed-up inner rim and the degree of outer diskflaring (which in these models is influenced by shadowingby the inner rim).

From the slope of the sub-mm/mm SED (and under theassumption that the disks are optically thin at these wave-lengths), it has been shown that HAe and late HBe objectswith smaller MIR flux excess have larger mid-plane dustgrains, perhaps implying an evolutionary sequence flared→ self-shadowed disks due to grain growth and settling; ormay alternatively indicate enhanced small-particle replen-ishment for the flared systems (Acke et al., 2004).

Clearly, the spectrally and spatially resolved disk obser-vations made possible by instruments such as VLTI/MIDIwill prove crucial towards further exploring these excitingprospects for linking the overall disk structure, the prop-erties of mid-plane and surface layer dust, and their timeevolution towards the planet-building phase (see also thechapter by Dominik et al.).

Fig. 3.— Predicted radial distribution of NIR (left) and MIR(right) light for typical disk models. DD04 (Dullemond and Do-minik, 2004) refers to a 2D radiative transfer and hydrostatic equi-librium Monte Carlo code featuring a puffed up inner rim; CG97to the gradually flaring model ofChiang and Goldreich1997).We note that the CG97 disk model can not self-consistently dealwith the inner disk once the temperature of the ”super-heated” dustlayer increases beyond the dust sublimation temperature. Thisonly effects the NIR light profiles and here the disk was artifi-cially truncated at 0.21 AU. The central star is assumed to be oftype A0 with M=2.5 M� and R = 2.0 R�. The upper panels showthe intensity profiles, and the lower panels the cumulative bright-ness contributions, normalized to 1. Three qualitatively differentdisk regions are indicated by shadowing. This figure illustratesthat NIR observations almost exclusively probe the hot inner rimof the circumstellar disks, while MIR observations are sensitiveto disk structure over the first dozen or so AU. (Credit: Roy vanBoekel).

3.2 Dust Mineralogy and Mixing

The MIR spectral region contains strong resonances ofabundant dust species, both O–rich (amorphous or crys-

Group II, "self−shadowed" Group I, "flaring"

[mag]

Group II: Flat−Shadowed Group I: Flaring

Fig. 4.—Relation between12.5 µm sizes (measured as half-lightradius) and infrared slope (measured by IRAS colors) for the firstHAeBe stars observed with MIDI on the VLTI. This size-colorrelation is consistent with expectations from the SED classificationof Meeus et al.(2001) into objects with flaredvs.flat outer disks.

talline silicates) and C–rich (Polycyclic Aromatic Hydro-carbons, or PAHs). Therefore, MIR spectroscopy of (op-tically thick) proto-planetary disks offers a rich diagnosticof the chemical composition and grain size of dust in theupper disk layers or “disk atmosphere”. At higher spec-tral resolution,R ∼ few 100, gas emission such as [NeII]lines and lines of the Hund and higher Hydrogen series canbe observed. It should be noted that these observations donot constrain the properties of large (> 2 − 4µm) grains,since these have little or no spectral structure at MIR wave-lengths. In the sub-micron and micron size range however,different silicate particles can be well distinguished on thebasis of the shape of their emission features. With spatiallyresolved observations, this type of dust processing can besearched at different spatial scales in the disk, rather thanfor the (unresolved) system as a whole.

Van Boekel et al.(2004) have demonstrated the power ofspectrally resolved MIR interferometry, by spatially resolv-ing three proto-planetary disks surrounding HAeBe starsacross the N-band. The correlated spectra measured by theinterferometer (VLTI/MIDI) correspond to disk regions at1–2 AU. By combining these with unresolved spectra, thespectrum corresponding to outer disk regions at 2–20 AUcan also be deduced. These observations have revealedradial gradients in dust crystallinity, particle size (graingrowth), and at least in one case (HD 142527) chemicalcomposition. These early results have revived the discus-sion of radial and out-of-the-plane mixing.

9

Interstellar particles observed in the direction of thegalactic center (Kemper et al., 2004) are mainly small amor-phous Olivine grains (Fe2−2xMg2xSiO4) with some admix-ture of Pyroxene grains (Fe1−xMgxSiO3). In our planetarysystem, comets like Halley, Levi or Hale-Bopp (Hanner etal., 1994;Crovisier et al., 1997) show the features of crys-talline Forsterite (an Mg-rich Olivine, x=1), in particular theconspicuous emission at11.3µm.

The VLTI/MIDI observations have revealed that crys-tallinization in young circumstellar disks can be a very ef-ficient process, by revealing a very high fraction of crys-talline dust in the central 1–2 AU; and that the outer 2–20 AU disk posseses a markedly lower crystalline fraction,but still much higher than that of the interstellar medium.Combined with the very young evolutionary status of thesesystems (as young as 1 Myr), these observations imply thatproto-planetary disks are highly crystaline well before theonset of planet formation. The presence of crystalline dustin relatively cold disk regions (or in Solar System comets)is surprising, given that temperatures in these regions arewell below the glass temperature of∼ 1000 K, but can beexplained by chemical models of proto-planetary disks thatinclude the effects of radial mixing and local processes inthe outer disk (Gail, 2004;Bockelee-Morvan et al., 2002).The radial gradient found in dust chemistry is also in qual-itative agreement with the predictions of these models. Fi-nally, the width of the silicate feature in the MIDI spectracorresponding to the inner and outer disk have revealed aradial gradient in (amorphous) grain size, with small grains(broader feature) being less abundant in the inner disk re-gions. This is perhaps expected, given that grain aggrega-tion is a strong function of density (see the chapter by Nattaet al.). Fig. 5 includes an example illustrating radial dustmineralogy gradients for the HAe object HD 144432.

These radial gradients in dust mineralogy are of coursenot restricted to HAeBe stars, but are expected in all YSOdisks. In Fig. 5 we show preliminary VLTI/MIDI resultsfor two low-mass objects of different evolutionary status,TW Hya and RY Tau (Leinert and Schegerer, 2006, inpreparation). For TW Hya, the spatially unresolved N-band spectrum (right panel) shows no strong evidence forcrystalline silicates, while the correlated flux spectrum (leftpanel) shows a weak but highly processed silicate band.Preliminary modelling indicates that the inner disk regionof TW Hya contains a high fraction of crystalline silicates(30%), and that the amorphous grains have aggregated tolarger units. As with the HAeBe stars, not all T Tauri ob-jects show the same features; e.g. the effect is qualitativelysimilar but quantitatively less pronounced in RY Tau, as canbe seen in Fig. 5.

4. OTHER PHENOMENA

4.1 Outflows and Winds

The power of spectrally resolved interferometric mea-surements has recently become available in the NIR spec-

Fig. 5.—Evidence for radial changes of dust composition in thecircumstellar disks of young stars. Infrared spectra of the inner≈ 1–2 AU are shown in the left panels, and of the total disk (≈ 1–20 AU) in the right panels. The different shapes of the silicateemission lines testify that in the inner disk – as probed by the in-terferometric measurement – the fraction of crystalline particlesis high while the fraction of small amorphous particles is signifi-cantly reduced. TW Hya (0.3 L�) and RY Tau (18 L�) approx-imately bracket the luminosity range of T Tauri stars, the 10 L�

HAe star HD 144432 (fromvan Boekel et al., 2004) is includedto emphasize the similarity between these two classes of youngobjects. Credit: Roy van Boekel.

10

tral range as well (VLTI/AMBER,Malbet et al., 2004;CHARA/MIRC, Monnier et al., 2004). In addition to pro-viding the detailed wavelength dependence of inner diskcontinuum (dust) emission, these capabilities enable de-tailed studies of the physical conditions and kinematics ofthe gaseous components in which emission and absorptionlines arise (e.g. Brγ , CO and H2 lines; as probes of hotwinds, disk rotation and outflows, respectively).

As a rather spectacular example of this potential, dur-ing its first commissioning observations the VLTI/AMBERinstrument spatially resolved the luminous HBe objectMWC 297, providing visibility amplitudes as a functionof wavelength at intermediate spectral resolutionR = 1500

across a2.0 − 2.2µm band, and in particular a finely sam-pled Brγ emission line (Malbet et al., 2006). The interfer-ometer visibilities in the Brγ line are∼ 30% lower thanthose of the nearby continuum, showing that the Brγ emit-ting region is significantly larger than the NIR continuumregion.

Known to be an outflow source (Drew et al., 1997),a preliminary model has been constructed byMalbet etal. (2006) in which a gas envelope, responsible for the Brγ

emission, surrounds an optically thick circumstellar disk(the characteristic size of the line emitting region being 40%larger than that of the NIR disk). This model is succes-ful at reproducing the new VLTI/AMBER measurements aswell as previous continuum interferometric measurementsat shorter and longer baselines (Millan-Gabet et al., 2001;Eisner et al., 2004), the SED, and the shapes of the Hα, Hβ

and Brγ emission lines.The precise nature of the MWC 297 wind however re-

mains unclear; the limited amount of data obtained in thesefirst observations can not, for example, discriminate be-tween a stellar or disk origin for the wind, or between com-peting models of disk winds (e.g.Casse and Ferreira, 2000;Shu et al., 1994). These key questions may however beaddressed in follow-up observations of this and similarobjects, perhaps exploiting enhanced spectral resolutionmodes (up toR = 10000 possible with VLTI/AMBER) andclosure phase capabilities.

Quirrenbach et al.(2006) have also presented prelimi-nary VLTI/MIDI results of rich spectral content for the windenvironment in another high luminosity YSO, MWC 349–A. The general shape of the N-band spectrally resolved vis-ibilities is consistent with disk emission, as in theLeinertet al. (2004) results (Section 3.1). In addition, the visibilityspectrum contains unprecedented richness, displaying overa dozen lines corresponding to wind emission in forbidden([NeII], [ArIII], [SIV]) and H recombination lines. The rel-ative amplitudes of the visibilities measured in the contin-uum and in the lines indicate that the forbidden line regionis larger than the dust disk, and that the recombination re-gion is smaller than the dust disk (at least along some di-rections). Moreover, differential phases measured with re-spect to the continuum also display structure as a functionof wavelength, showing clear phase shifts that encode infor-mation about the relative locations of the emitting regions

for the strongest emission lines.Substantial modelling efforts is required to fully exploit

the multitude of observables provided by, ideally, the com-bination of these new-generation instruments. It is to beexpected that the resulting new insight will play a key rolein elucidating the interplay of outflow and disk phenomenain these luminous young stars.

4.2 Multiplicity and Stellar Masses

Most young (and main sequence) stars are members ofmultiple systems, likely as a result of the star formationprocess itself (see the chapters by Goodwin et al. and byDuchene et al.). The measurement of the physical orbitsof stars in multiple systems provide the only direct methodfor measuring stellar masses, a fundamental stellar param-eter. In turn, by placing stars with measured masses in anHR-diagram, models of stellar structure and evolution canbe critically tested and refined, provided the mass measure-ments have sufficient accuracy (see the chapter by Math-ieu et al.). Generally speaking, dynamical and predictedmasses agree well for1−10M� stars in the main sequence.However, fundamental stellar properties are much less wellknown for pre-main sequence (PMS) stars, particularly oflow-mass (Hillenbrand and White, 2004), and call for massmeasurements with better than 10% accuracy.

Optical interferometers can spatially resolve close bina-ries (having separations of order∼ 1 mas, or 1/10ths AU at150 pc), and establish the apparent astrometric orbits, mostnotably its inclination (for eclipsing binaries, the inclinationis naturally well constrained). In combination with radialvelocity measurements using Doppler spectroscopy, a fullsolution for the physical orbit and the properties of the sys-tem can then be obtained, in particular the individual stel-lar masses and luminosities. This method has proved veryfruitful for critical tests of stellar models for non-PMS stars(e.g. Boden et al., 2005), however few simultaneous radialvelocity and astrometric measurements exist for PMS stars.

Boden et al.(2005b) performed the first direct measure-ment of PMS stellar masses using optical interferometry,for the the double-lined system HD 98800–B (a pair ina quadrupole system located in the TW Hya association).Using observations made at the KI and by the Fine Guid-ance Sensors aboard the Hubble Space Telescope, and incombination with radial velocity measurements, these au-thors establish a preliminary orbit which allowed determi-nation of the (sub-solar) masses of the individual compo-nents with 8% accuracy. Comparison with stellar modelsindicate the need for sub-solar abundances for both com-ponents, although stringent tests of competing models willonly become possible when more observations improve theorbital phase coverage and thus the accuracy of the stellarmasses derived.

Other PMS systems being pursued for which accuratemasses may be obtained include: MWC 361–A at the highmass end of the spectrum (likely two Be stars,Millan-Gabetet al., 2001;Monnier et al., 2006, in preparation); Haro 1–

11

14C (Schaefer et al., 2005; and see also the chapter byMathieu et al.); HD 98800–A (Boden et al., 2005); and thetriple T Tauri system GW Ori (also believed to contain cir-cumstellar and circumbinary disks, and for whichBergeret al. (2006) present preliminary separation vectors and thefirst reconstructed image of a young multiple system).

Naturally, disk phenomena (circumstellar and/or cir-cumbinary) are also often observed in young multiple sys-tems (see the chapter by Monin et al.); and a number ofrecent observations also address this interesting and morecomplicated situation. Indeed, successful modelling of theindividual HD 98800–B components byBoden et al.(2005)requires a small amount of extinction (AV = 0.3) toward theB–components, not required toward the A–components,and possibly due to obscuration by circumbinary disk ma-terial around the B system (originally hypothesized byTokovinin, 1999). As another example, based on a low-leveloscillation in the visibility amplitude signature in the PTIdata for FU Ori,Malbet et al.(2005) claim the detectionof an off-centered spot embedded in the disk that could bephysically interpreted as a young stellar or proto-planetarycompanion (located at∼ 10 AU), and possibly be at theorigin of the FU Ori outburst itself.

In summary, the improved infrared sensitivity of new in-terferometer facilities has enabled the first measurementsof PMS binary orbits and stellar masses, and promises toresult in establishing fundamental stellar properties foralarger sample of the closest systems across the mass range,and in critical tests of state of the art stellar models. Asnoted byMathieu(1994), spectroscopic detection of radialvelocity changes is challenging in the presence of strongemission lines, extreme veiling or rapid rotation, all ofwhich are common in PMS stars. Therefore, imaging tech-niques are unique discovery tools for PMS multiples. Forthe closest systems in particular, interferometric techniquescan reveal companions with separations∼ ×10 smallerthan single-aperture techniques (e.g. Speckle or AdaptiveOptics). Therefore, the increasing number of new detec-tions will add an important new sample to statistical stud-ies aimed at determining the multiplicty fraction for youngstars, its dependence on separation, stellar properties andevolutionary status, and implications for circumstellar disksurvival in close binary environments.

5. SUMMARY AND OPEN QUESTIONS

Summary

Long-baseline interferometers operating at near andmid-infrared wavelengths have spatially resolved the cir-cumstellar emission of a large number of YSOs of varioustypes (60 objects published to date), most within the lastfew years. While the new observables are relatively basicin most cases (broadband visibility amplitudes), they haveprovided fundamentally new direct information (character-istic sizes and crude geometry), placing powerful new con-straints on the nature of these circumstellar environments.

In addition, more powerful observational capabilities com-bining high spatial resolution and spectral resolution havejust begun to deliver their potential as unique tools to studyYSO environments in exquisite detail and to provide break-through new molecular and kinematic data.

The principal well-established observational facts maybe summarized as follows:

− For HAe, late HBe and T Tauri objects the measuredNIR characteristic sizes are much larger (by factorsof ∼ 3 − 7) than predicted by previous-generationmodels which had been reasonably successful at re-producing the unresolved spectro-photometry. In par-ticular, the measured sizes are larger than initial in-ner dust hole radii estimates and than magnetospherictruncation radii. The measured NIR sizes stronglycorrelate with the central luminosity, and the empiri-cal NIR sizevs. luminosity relation (Fig. 2) suggeststhat the NIR emission is located at radii correspond-ing to sublimation radii for dust directly heated bythe central star. These measurements have motivatedin part the development of a new class of models forthe origin of the NIR disk emission (the “puffed-up”inner wall models) which not only provide qualita-tive agreement with the interferometer data, but alsosolves the SED “NIR bump” problem for HAe ob-jects.

− Some (but not all) of the earliest-type HBe objectsare different than the later types in terms of theirNIR size-scales. Many are inferred to have relativelysmall inner dust holes in better agreement with theclassical disk picture, while others are have largerNIR sizes consistent with the inner rim models (orperhaps even larger than predictions of either com-peting disk model, e.g. LkHα 101).

− FU Ori, the prototype for the sub-class of YSOS ex-pected to have disks which dominate the total lumi-nosity, is well described by the canonical model tem-perature law (based on modelling the NIR interfer-ometry data and SED).

− Young disks have also been resolved at MIR wave-lengths, revealing characteristic sizes which correlatewith red color (IRAS12− 25µm). This link appearsto support an SED-based classification based on thedegree of outer disk flaring, although the current sam-ple of measured MIR sizes is too small to extract afirm conclusion.

− Spectrally-resolved MIR visibilities have emerged asa powerful tool to probe the dust minerology in youngdisks. Silicate dust appears to be more crystalline(less amorphous) close to the central star than in thedisk as a whole.

− The power of spectrally-resolved visibilities, bothNIR and MIR, has also been dramatically demon-strated in observations of lines corresponding to the

12

gas component in wind phenomena associated withhigh-mass young stars.

− Preliminary masses have been measured for youngstars for the first time using interferometric tech-niques, in combination with spectroscopic radial ve-locity techniques, and a number of systems are be-ing pursued with the primary goal of providing usefulconstraints to state of the art models of stellar struc-ture and evolution.

Open Questions

The new interferometric observations have allowed a di-rect view of the inner regions of YSO accretion disks forthe first time. While this work has motivated significantadditions to our standard picture of these regions (such asthe puffed-up inner rim and spatially-varying dust proper-ties), continued theoretical development will require moresophisticated and complete interferometry data. In this sec-tion, we outline the most pressing un-resolved issues ex-posed by the progress describe above:

− A disk origin for the resolved NIR emission has notbeen unambiguously established. Most notably,Vin-covik et al.(2006) (and references therein) considerthat the remaining scatter in the size-luminosity rela-tion for HAeBe objects indicates that the bright-ringmodel does not adequately describe all the objectswith a uniform set of parameters, and propose insteada disk plus compact halo model, where the NIR emis-sion is in fact dominated by the halo component. Be-cause the halo need not be spherically symmetric, theelongations detected at the PTI do not rule it out. Ef-fectively re-igniting thedisk vs. envelopedebate ofthe previous decade (see e.g.Natta et al., 2000),Vin-covik et al.(2006) also show that this model equiva-lently reproduces the SED, including the NIR bump.

− The detailed properties of the putative puffed-up in-ner rim remain to be established. Assumptions aboutthe grain size and density directly impact the ex-pected rim location (dust sublimation radius), andbased on size comparisons (Fig. 2) current observa-tions appear to indicate that a range of properties maybe needed to explain all disks, particularly among theT Tauri objects. HAe objects have NIR sizes consis-tent with optically thin dust and relatively large grains(∼ 1µm). The lowest luminosity T Tauri objectshowever favor either optically thick dust or smallergrains, or a combination of both. Currently however,the T Tauri sample is still small, several objects arebarely resolved, and part of the scatter could be dueto unreliable SED decomposition due to photometricvariability and uncertain stellar properties. Detailedmodelling of individual objects and crucial support-ing observations such as infrared veiling (to directlyestablish the excess flux) are needed to help resolvethese uncertainties.

− Using IR spectroscopy, it has also been establishedthat thegasin young disk systems extends inward ofthe interferometrically deduced inner dust radii (seethe chapter by Najita et al.). We note however thatcurrently the interpretation of the CO spectroscopydoes not take into account the inner rim, relying in-stead on previous-generation models that have beenall but ruled out for the inner few AU.

− The dust and gas disks do not terminate at the sameradius, and the dust disk does not terminate at themagnetospheric accretion radius. Whether or notmagnetospheric accretion theories can accomodatethese observations remains to be explored.

− The effects of NIR emission by gas in the inner diskand of NIR scattering by large scale dust, both on themeasured visibilities and on the SED decomposition(both of which affect the inferred NIR sizes), need tobe quantified in detail on a case-by-case basis.

− Large scale (∼ 50 − 1000 mas) “halos” contributingsmall but non-negligible NIR flux (∼ 10%) are ofteninvoked as an additional component needed to satis-factorily model some YSO disks (e.g. the FU Orio-nis objects, or most recently in the IOTA-3T HAeBesample ofMonnier et al., 2006, in preparation). Theorigin of this halo material is however unclear, as isits possible connection with the compact halos pro-posed byVincovik et al.(2006). The observationalevidence should motivate further work on such multi-component models.

− Late (HAe,late HBe) and early HBe systems ap-pear to have different NIR scale properties (see Sec-tion 2.1), although there are notable exceptions, andthe sample is small. Differences between these typeshave also been noted in terms of their mm emission(absent in most early HBe, in contrast to the late Her-big and T Tauri objects), perhaps due to outer-diskphotoevaporation (Hollenbach et al., 2000, and seealso the chapter by Dullemond et al.). Whether theobserved differences in inner disk structure are dueto different accretion mechanisms (disk accretion vs.magnetospheric accretion) (see alsoVink et al, 2005)or to gas opacity effects remains to be investigated.

− Do all FU Orionis objects conform to the canonicalaccretion disk model? If so, the additional questionarises as to why their temperature structure would bethat simple. Their light curves are fading, indicatinga departure from a steady-state disk. Indeed, the highoutburst accretion rates are expected to decay, and todo so in a radially dependent manner, leading to non-standard temperature laws. The new KI data of threeadditional FU Ori objects indicate that we have moreto learn about accretion around these enigmatic ob-jects, and first observations using VLTI/MIDI do notindicate a simple connection between the NIR and

13

MIR radial disk structure. Indeed, fitting the MIRvisibilities and SED of V1647 Ori requires a very flatq = −0.5 radial temperature exponent (Abraham etal., 2006), and preliminary analysis of FU Ori itself(Quanz, 2006) indicates that the model constructedby Malbet et al.(2005) to fit the NIR AMBER dataand the SED does not reproduce (over-estimates) themeasured N-band visibilities.

− The overall disk structure needs to be secured. Whilethe NIR data probe primarily the hottest dust, theMIR observations are sensitive to a wider range oftemperatures. Given the complexity of the disk struc-ture under consideration, correctly interpreting theMIR and NIR data for the same object is provingchallenging. Assumptions made for modelling theNIR data, the MIR dust features, and CO line pro-files are generally not consistent with each other. Ex-citingly, the joint modeling of near and mid-IR highresolution data can yield the temperature profile ofthe disk surface layers.

− Chemical models of proto-planetary disks, includ-ing the effects of radial transport and vertical mix-ing, must explain the observed dust mineralogy, andthe differences seen among different types of objects.Also, we note that the effect of the inner rim (whichcontributes∼ 20% of the MIR flux – see Fig. 3)on the interpretation of the observed Silicate featuresand inferred mineralogy gradients must be assesed indetail.

6. FUTURE PROSPECTS

This review marks the maturation of two-telescope(single-baseline) observations using infrared interferome-try. However, much more information is needed to seriouslyconstrain the next generation of models (and the prolifer-ating parameters) and to provide reliable density and tem-perature profiles as initial conditions for planet formationtheories. Fortunately, progress with modern interferome-ters continues to accelerate and will provide the unique newobservations that are needed. In this section, we brieflydiscuss the expected scientific impact on yet-unexploredareas that can be expected from longer baselines, multi-wavelength high-resolution data, spectral line and polariza-tion capabilities, closure phase data and aperture synthesisimaging. We will end with (humble) suggestions for diskmodelers who wish to prepare for the new kinds of interfer-ometric data in the pipeline.

Detailed Disk Structure

All interferometric observations of YSO disks to datehave used baselines. 100 meters, corresponding to angularresolution of∼ 2 and 11 milliarcsec at 2.2 and 10 micronsrespectively. While sufficient to resolve the overall extent

of the disks, longer baselines are needed to probe the inter-nal structure of this hot dust emission. For instance, currentdata can not simultaneously constrain the inner radius andthickness of the inner rim dust emission, the latter often as-sumed to be “thin” (10 − 25% of radius). Nor can currentdata independently determine the fraction of light from thedisk compared to star (critical input to the models), relyinginstead on SED decomposition.

Longer baselines from the CHARA Interferometer(330 meters) and from VLTI Interferometer (200 meters)can significantly increase the angular resolution and allowthe rim emission to be probed in detail. For instance, if theNIR emission is contained in a thin ring, as expected for“hot inner wall” models, there will be a dramatic signal inthe second lobe of the interferometer visibility response.Competing “halo” models predict smoother fall-offs inbrightness (and visibility) and thus new long-baseline datawill provide further definitive and completely unique con-straints on the inner disk structure of YSOs (first long-baseline CHARA observations of YSOs were presented byMonnier et al., 2005b).

While even two-telescope visibility data provide criti-cal constraints, ultimately we strive for aperture synthesisimaging, as routinely done at radio wavelengths. The firststeps have recently been taken with closure phase resultson HAeBe objects using IOTA/IONIC3 (Millan-Gabet etal., 2006b;Monnier et al., 2006; both in preparation). Clo-sure phases are produced by combining light from threeor more telescopes, and allow “phase” information to bemeasured despite atmospheric turbulence (see e.g.Mon-nier, 2000). While this phase information is essential forimage reconstruction, the closure phase gives unambiguoussignal of “skewed” emission, such as would arise from aflared disk viewed away from the pole. “Skew” in this con-text indicates a deviation from a centro-symmetric bright-ness.

Closure phases are interesting for measuring the verticalstructure of disks. Early hot inner wall models (e.g.,Dulle-mond et al., 2001) posited vertical inner walls. Recently,the view was made more realistic byIsella and Natta(2005)by incorporating pressure-dependent dust sublimation tem-peratures to curve the inner rim away from the midplane.Closure phases can easily distinguish between these scenar-ios because vertical walls impose strong skewed emissionwhen a disk is viewed at intermediate inclination angles,while curved inner walls appear more symmetric on thesky unless viewed nearly edge-on. NIR closure phase datawill soon be available from IOTA/IONIC3, VLTI/AMBERand CHARA/MIRC that can be used to measure the curva-ture and inner rim height through model fitting of specificsources.

Aperture synthesis imaging of selected YSOs using theVLTI and CHARA arrays should also be possible withinthe next five years. By collecting dozens of closure phasetriangles and hundreds of visibilities, simple images can becreated as has been demonstrated for LkHα 101 using theaperture masking technique. These images will allow the

14

first model-independent tests of disk theories. All currentinterpretations are wed to specific (albeit general) modelsand the first images may be eye-opening.

As described earlier (Section 3), it is essential to knowthe properties of circumstellar dust particles in YSO disksin order to interpret a broad range of observations, most es-pecially in standard SED modelling. Interferometers canmeasure dust properties, such as size distribution and com-position, using the combination of NIR, MIR, and polar-ization measurements. Efforts are being explored at thePTI, IOTA, VLTI and CHARA interferometers to measurethe disk emission in linearly polarized light (seeIrelandet al., 2005 for similar first results for dust around AGBstars). Because scattering is heavily dependent on grain-size, the polarization-dependent size estimates at differentwavelengths will pinpoint the sizes of the grains present inthe inner disk.

With its Nulling capability at MIR wavelengths, the KI(Serabyn et al., 2004) will soon address a wide range ofYSO phenomena, from spectrally resolved MIR disk struc-ture in young disks to exo-zodiacal emission (Serabyn etal., 2000), the latter being a crucial element in the selec-tion of favorable targets for future space-based planet find-ing missions (e.g. Terrestrial Planet Finder/Darwin,Frid-lund, 2003).

Proto-planets forming in circumstellar disks are ex-pected to carve fine structure, such as gaps, opening thepossibility of directly detecting this process via high dy-namic range interferometric techniques (e.g. highly accu-rate visibility amplitude measurements). While some initialsimulations are pessimistic that new interferometers canresolve such disk structures (Wolf et al., 2002), it is clearthat theorists have only begun to investigate the many pos-sibilities and interferometers are gearing up to meet thenecessary observational challenges ahead.

Finally, although outside the scope of this review, wepoint out the tremendous potential of the upcoming Ata-cama Large Millimiter Array (ALMA,Wootten, 2003). Inparticular, the combination of infrared, mm and sub-mm in-terferometry will probe the entire YSO disk at all scales.

Dynamics

New spectral line (e.g. VLTI/AMBER) capabilities haveexciting applications. For instance, spectro-interferometrycan measure the Keplerian rotation curves (in CO) for disksto derive dynamical masses of young stars as has been doneusing mm-wave interferometry (Simon et al., 2000). As ageneral statement, the potentially very powerful combina-tion of spectroscopy and interferometry has only begun tobe explored.

The dynamic time scale for inner disk material can be. 1 year, thus we might expect to see changes with time.The higher the angular resolution, the greater the possibil-ity that we can identify temporal changes in the circum-stellar structures. If disk inhomogeneities are present, newmeasurements will discover them and track their orbital mo-tions. Evidence of inner disk dynamics has been previously

inferred from reflection nebulosity (e.g. HH 30,Stapelfeldtet al., 1999) and photometric (e.g.Hamilton et al., 2001)variability; but only recently directly imaged at AU scales(LkHa 101,Tuthill et al., 2002).

The Star - Disk - Outflow Connection

The interface between the star, or its magnetosphere, andthe circumstellar disk holds the observational key into howdisks mediate angular momentum transfer as stars accretematerial. The relevant spatial scales (1 to fewR?) maybe resolvable by the longest baselines available, providingunique tests of magnetospheric accretion theories (see thechapter by Bouvier et al.).

In addition to being surrounded by pre-planetary disks,YSOs are often associated with outflow phenomena, includ-ing optical jets, almost certainly associated with the diskac-cretion process itself (see the chapter by Bally, Reipurth andDavis). The precise origin of the jet phenomena is howeverunknown, as is the physical relation between the young star,the disk, and jets (see the chapter by Ray et al.). Discrim-inating among competing models, e.g jets driven by exter-nal large scale magnetic fields (the “disk wind” models orstellar fields models) or by transient local disk fields, canbe helped by determining whether the jets launch near thestar, or near the disk. Observationally, progress depends inpart on attaining sufficient spatial resolution to probe thejetlaunching region directly, a natural task for optical interfer-ometers (e.g.Thiebaut et al., 2003).

New Objects

Essentially unexplored areas concern types of objects forwhich little or no observations have been carried out to date.We point out here important astrophysical questions con-cerning types of objects for which the observational efforthas just begun, or can be expected to flourish in the next fewyears.

Infrared interferometry can directly address importantquestions concerning the early evolution of high mass (O–B) protostars (see the chapter by Cesaroni et al. for a re-view of the pressing issues), and observational work in thisarea using infrared interferometry has just begun (with theVLTI/MIDI observations ofQuirrenbach et al., 2006;Feldtet al., 2006). Most notably, it is not presently known whatrole, if any, accretion via circumstellar disks plays in theformation of high-mass stars. If present, interferometerscan resolve circumstellar emission, and determine whetheror not it corresponds to a circumstellar disk. Further, bystudying disk properties as a function of stellar spectraltype, formation timescales and disk lifetimes can be con-strained.

Current studies have mainly targeted young disk sys-tems, i.e. class II (Adams et al., 1987). More evolved,class III, disks have remained thus far unexplored (the onlyexception being V380 Ori inAkeson et al., 2005b), dueto the sensitivity limitations that affected the first obser-vations. Moreover, directly establishing the detailed diskstructure of transition objects in the planet-building phase

15

and of debris disks on sub-AU to 10s AU scales will likelybe an energetic area of investigation as the field progressestowards higher dynamic range and spatial frequency cover-age capabilities. First steps have been taken in this direc-tion, with NIR observations that resolve the transition ob-ject TW Hya (Eisner et al., 2006) and detect dust in the de-bris disk around Vega (Ciardi et al., 2001;Absil et al., 2006,in preparation).

Finally, we note that prospects also exist for the directdetection of exo-planets from the ground, including inter-ferometric techniques, and we refer the reader to the chapterby Beuzit et al.

Comments on Modelling

A number of new physical ingredients must be intro-duced into current disk models in order to take advantageof interferometer data. Disk models should take into ac-count both vertical variations (settling) and radial dust dif-ferences (processing and transport). Also, the structure ofinner wall itself depends on the physics of dust destruction,and may depend on gas cooling and other physical effectsnot usually explicitly included in current codes. Perhapscodes will need to calculate separate temperatures for eachgrain size and type to accurately determine the inner rimstructure. Lastly, we stress that the many free parameters inthese models can only be meaningfully constrained throughan aggressive data-driven approach, fitting not only SEDs(normal situation) but NIR and MIR interferometry simul-taneously for many individual sources. With these newmodels, disk mass profiles and midplane physical condi-tions will be known, providing crucial information bridgingthe late stages of star formation to the initial conditions ofplanet formation.

Acknowledgments.We thank Roy van Boekel for providing Figs. 3 and 5

and for helpful discussions. We also thank many of ourcolleagues for illuminating discussions while preparing thisreview, in particular Myriam Benisty, Jean-Philippe Berger,Andy Boden, Kees Dullemond, Josh Eisner, Lynne Hillen-brand and Andreas Quirrenbach.

REFERENCES

Abraham P. et al. (2004)Astron. Astrophys., 428, 89-97.Abraham P. et al. (2006)Astron. Astrophys., in press.Acke B. et al. (2004)Astron. Astrophys., 422, 621-626.Adams F. C., Lada C. J. and Shu F. H. (1987)Astrophys. J., 312,

788-806.Akeson R. L. et al. (2000)Astrophys. J., 543, 313-317.Akeson R. L., Ciardi D. R., van Belle G. T. and Creech-Eakman,

M. J. (2002)Astrophys. J., 566, 1124-1131.Akeson R. L. et al. (2005)Astrophys. J., 622, 440-450.Akeson R. L. et al. (2005b)Astrophys. J., 635, 1173-1181.Berger J.-P. et al. (2005) inProtostars and Planets V Poster Pro-

ceedingshttp://www.lpi.usra.edu/meetings/ppv2005/pdf/8398.pdf

Bjorkman J. E. and Wood K. (2001)Astrophys. J., 554, 615-623.Boden A. F., Torres G. and Hummel C. A. (2005)Astrophys. J.,

627, 464-476Boden A. F., Torres G, and Hummel C. A. (2005b)Astrophys. J.,

627, 464-476.Bockelee–Morvan D. et al. (2002)Astron. Astrophys., 384, 1107-

1118.Casse F. and Ferreira J. (2000)Astron. Astrophys., 353, 1115-

1128.Chiang E. and Goldreich P. (1997)Astrophys. J., 490, 368.Ciardi D. R. et al. (2001),Astrophys. J., 559, 2, 1147-1154.Colavita M. M. et al. (1999)Astrophys. J., 510, 505-521.Colavita M. M. et al. (2003)Astrophys. J., 529, L83-L86.Colavita M. M., Wizinowich P. L. and Akeson R. L. (2004) in

Proceedings of the SPIE, 5491, 454.Clarke C. J., Gendrin A. and Sotomayor M. (2001)Mon. Not. R.

Astron. Soc.328, 485-491.Crovisier J. et al. (1997)Science, 275, 1904-1907.D’Alessio P., Calvet N., Hartmann L., Muzerolle J. and Sitko M.

(2004) in Star Formation at High Angular Resolution, IAUSymposium, (M. G. Burton, R. Jayawardhana and T. L. Bourke,eds.),221, 403-410.

Danchi W. C., Tuthill P. G. and Monnier J. D. (2001)Astrophys.J., 562, 440-445.

Dominik C., Dullemond C. P., Waters L. B. F. M. and Walch S.(2003)Astron. Astrophys., 398, 607-619.

Dullemond C. P., Dominik C. and Natta A. (2001)Astrophys. J.,560, 957-969.

Dullemond C. P. and Dominik C. (2004)Astron. Astrophys., 417,159-168.

Drew J. et al. (1997)Mon. Not. R. Astron. Soc., 286, 538-548.Eisner J. A. et al. (2003)Astrophys. J., 588, 360-372.Eisner J. A. et al. (2004)Astrophys. J., 613, 1049-1071.Eisner J. A. et al. (2005)Astrophys. J., 623, 952-966.Eisner J. A., Chiang, E. I. and Hillenbrand, L. A. (2006)Astro-

phys. J., in press.Feldt M. et al. (2006) inThe Power of Optical / IR Interferometry:

Recent Scientific Results and 2nd Generation VLTI Instrumen-tation, (F. Paresce, A. Richichi, A. Chelli and F. Delplancke,eds.), ESO Astrophysics Symposia, Springer-Verlag, Garch-ing.

Fridlund M. C. (2003) inProceedings of the SPIE, 5491, 227.Gail H.-P. (2004)Astron. Astrophys., 413, 571-591.The Very Large Telescope Interferometer – Challenges for the Fu-

ture (2004), (Garcia P. J. V., Glindeman A., Henning T. andMalbet F. eds.), Astrophysics and Space Science Volume 286,Nos. 1-2, Kluwer Academic Publishers, Dordrecht.

Gil C. et al. (2006) inThe Power of Optical / IR Interferometry:Recent Scientific Results and 2nd Generation VLTI Instrumen-tation, (F. Paresce, A. Richichi, A. Chelli and F. Delplancke,eds.), ESO Astrophysics Symposia, Springer-Verlag, Garch-ing.

Hamilton C. et al. (2001)Astrophys. J., 554, L201-L204.Hanner M. S., Lynch D. K. and Russell R. W. (1994)Astrophys.

J., 425, 274-285.Hale D. et al. (2000)Astrophys. J., 537, 998-1012.Hartmann L. and Kenyon S. J. (1985)Astrophys. J., 299, 462-478.Hartmann L. and Kenyon S. J. (1996)Annu. Rev. Astron. Astro-

phys., 34, 207-240.Hartmann L., Calvet N., Gullbring E. and D’Alessio P. (1998)As-

trophys. J., 495, 385.Herbig G. H., Petrov P. P. and Duemmler R. (2003)Astrophys. J.,

16

595, 384-411.Hillenbrand L. A., Strom S. E., Vrba F. J., and Keene J. (1992).

Astrophys. J., 397, 613-643.Hillenbrand L. A. and White R. J. (2004)Astrophys. J., 604, 741-

757.Hinz P. M., Hoffmann W. F. and Hora J. L. (2001)Astrophys. J.,

561, L131-L134.Hollenbach D., Yorke H. W. and Johnstone D. (2000) inProtostars

and Planets IV, (Mannings, V., Boss, A. P. and Russell, S. S.eds.), pp. 401-428, University of Arizona Press, Tucson.

Ireland M., Tuthill P. G., Davis J. and Tango W. (2005)Mon. Not.R. Astron. Soc., 361, 337-344.

Isella A. and Natta A. (2005)Astron. Astrophys., 438, 899-907.Jayawardhana R. et al. (1998)Astrophys. J., 503, L79.Johns-Krull C. M. and Valenti V. A. (2001)Astrophys. J., 561,

1060-1073.Kemper F., Vriend W. J. and Tielens A. G. G. M. (2004)Astrophys.

J., 609, 826-837.Kenyon S. J. and Hartmann L. W. (1987)Astrophys. J., 323, 714-

733.Kenyon S. J. and Hartmann L. W. (1991)Astrophys. J., 383, 664-

673.Koerner D. W., Ressler M. E., Werner M. W. and Backman D. E.

(1998)Astrophys. J., 503, L83.Kuchner M. and Lecar M. (2002)Astrophys. J., 574, L87-L89.Lachaume R., Malbet F. and Monin J.-L. (2003)Astron. Astro-

phys., 400, 185-202.Principles of Long Baseline Interferometry(2000) (Lawson P.

ed.), published by National Aeronautics and Space Adminis-tration, Jet Propulsion Labratory, California Institute of Tech-nology, Pasadena.

Leinert Ch., Haas M., Abraham P. and Richichi A. (2001)Astron.Astrophys., 375, 927.

Leinert Ch. (2004) inProceedings of the SPIE, 5491, 19.Leinert Ch. et al. (2004)Astron. Astrophys., 423, 537-548.Lin D. N. C., Bodenheimer P. and Richardson D. C. (1996) Nature,

380, 606-607.Liu W. M. et al. (2003)Astrophys. J., 598, L111-L114.Liu W. M. et al. (2005)Astrophys. J., 618, L133-L136.Lynden-Bell D. and Pringle J. E. (1974)Mon. Not. R. Astron. Soc.,

168, 603-637.Malbet F. and Bertout C. (1995)Astron. Astrophys. Suppl., 113,

369.Malbet F. et al. (1998)Astrophys. J., 507, L149-L152.Malbet F. et al. (2004) inProceedings of the SPIE, 5491, 1722.Malbet F. et al. (2005)Astron. Astrophys., 437, 627-636.Malbet F. et al. (2006)Astron. Astrophys., in press.Mathieu R. D. (1994)Annu. Rev. Astron. Astrophys., 32, 465-530.McCabe C., Duchene G. and Ghez A. (2003)Astrophys. J., 588,

2, L113-L116.Meeus G. et al. (2001)Astron. Astrophys., 365, 476-490.Millan-Gabet R., Schloerb F. P., Traub W. A., Malbet F., Berger J.-

P., and Bregman, J. D. (1999)Astrophys. J., 513, L131-L143.Millan-Gabet R., Schloerb F. P., and Traub W. A. (2001)Astro-

phys. J., 546, 358-381.Millan-Gabet R. et al. (2006)Astrophys. J., in press.Monnier J. D. (2000) inPrinciples of Long Baseline Interferome-

try (Lawson P. ed.), pp. 203-226, published by National Aero-nautics and Space Administration, Jet Propulsion Labratory,California Institute of Technology, Pasadena.

Monnier J. D. and Millan-Gabet R. (2002)Astrophys. J., 579, 694-698.

Monnier J. D. (2003) inReports on Progress in Physics, Vol. 66,Num. 5, 789-897

Monnier J. D. et al. (2004) inProceedings of the SPIE, 5491,1370.

Monnier J. D., Tuthill P. G., Ireland M. J., Cohen R. and Tan-nirkulam A. (2004b), inAmerican Astronomical Society Meet-ing 2005, 17.15.

Monnier J. D. et al. (2005)Astrophys. J., 624, 832-840.Monnier J. D. et al. (2005b) inProtostars and Planets V Poster

Proceedingshttp://www.lpi.usra.edu/meetings/ppv2005/pdf/8238.pdf

Muzerolle J., Calvet N., Hartmann L. and D’Alessio P. (2003)As-trophys. J., 597, L149-L152

Muzerolle J., D’Alessio P., Calvet N., Hartmann L. (2004)Astro-phys. J., 617, 406-417.

Natta A., Grinin V. and Mannings V. (2000) inProtostars andPlanets IV, (Mannings, V., Boss, A. P. and Russell, S. S. eds.),pp. 559-588, University of Arizona Press, Tucson.

Natta A. et al. (2001)Astron. Astrophys., 371, 186-197.The Power of Optical / IR Interferometry: Recent Scientific Results

and 2nd Generation VLTI Instrumentation(2006), (F. Paresce,A. Richichi, A. Chelli and F. Delplancke, eds.), ESO Astro-physics Symposia, Springer-Verlag, Garching.

Observing with the VLTI(2003) (Perrin G. and Malbet F. eds.),EAS Publications Series,Volume 6.

Quanz S. P. (2006) inThe Power of Optical / IR Interferometry:Recent Scientific Results and 2nd Generation VLTI Instrumen-tation, (F. Paresce, A. Richichi, A. Chelli and F. Delplancke,eds.), ESO Astrophysics Symposia, Springer-Verlag, Garch-ing.

Quirrenbach A. (2001),Annu. Rev. Astron. Astrophys., 39, 353-401.

Quirrenbach A., Albrecht S. and Tubbs R. N. (2006) inStarswith the B[e] phenomenon, (Michaela Kraus and Anatoly S.Miroshnichenko, eds.), ASP Conference Series,in press.

Rice W. K. M. et al. (2003)Mon. Not. R. Astron. Soc., 342, 79-85.Serabyn E., Colavita M. M. and Beichman C. A. (2000) in Pro-

ceedings of a meeting held at The Lunar and Planetary Insti-tute, 28–30 April, 1999, (Michael L. Sitko, Ann L. Sprague,and David K. Lynch, eds.), Astronomical Society of the PacificConference SeriesVol. 196, 357-365.

Serabyn E. et al. (2004) inProceedings of the SPIE, 5491, 806.Schaefer G. H., Simon M. and Prato L. (2005) inThe Power of

Optical / IR Interferometry: Recent Scientific Results and 2ndGeneration VLTI Instrumentation, (F. Paresce, A. Richichi, A.Chelli and F. Delplancke, eds.), ESO Astrophysics Symposia,Springer-Verlag, Garching.

Shu F. et al. (1994)Astrophys. J., 429, 781-807.Simon M., Dutrey A. and Guilloteau S. (2000)Astrophys. J., 545,

1034-1043.Stapelfeldt K. et al. (1999)Astrophys. J., 516, 2, L95-L98.ten Brummelaar T. A. et al. (2005)Astrophys. J., 628, 453-465.Tokovinin, A. (1999)Astron. Lett., 25, 669-671.Traub W. A. et al. (2004) inProceedings of the SPIE, 5491, 482.Tuthill P., Monnier J. D. and Danchi, W. C. (2001) Nature,409,

1012-1014.Tuthill P., Monnier J. D., Danchi W. C., Hale D. D. S. and Townes

C. H. (2002)Astrophys. J., 577, 826-838.van Boekel R. et al. (2004) Nature,432, 479-482.van Boekel R., Dullemond C. P. and Domink C. (2005)Astron.

Astrophys., 441, 563-571.Vinkovic D., Ivezic Z., Jurkic T. and Elitzur M. (2006)Astrophys.

17

J., 636, 348-361.Wilkin F. P. and Akeson R. L. (2003)Astrophys. Space Sci., 286,

145-150.Wootten A. (2003) inProceedings of the SPIE, 4837, 110.

18