Serb. Astron. J. } 183 (2011), 1 - 35 UDC 520.36–14DOI: 10.2298/SAJ1183001J Invited review

ASTRONOMICAL OPTICAL INTERFEROMETRY.II. ASTROPHYSICAL RESULTS

S. Jankov

Astronomical Observatory, Volgina 7, 11060 Belgrade 38, SerbiaE–mail: sjankov@aob.rs

(Received: November 24, 2011; Accepted: November 24, 2011)

SUMMARY: Optical interferometry is entering a new age with several ground-based long-baseline observatories now making observations of unprecedented spatialresolution. Based on a great leap forward in the quality and quantity of interfer-ometric data, the astrophysical applications are not limited anymore to classicalsubjects, such as determination of fundamental properties of stars; namely, theireffective temperatures, radii, luminosities and masses, but the present rapid devel-opment in this field allowed to move to a situation where optical interferometry is ageneral tool in studies of many astrophysical phenomena. Particularly, the adventof long-baseline interferometers making use of very large pupils has opened the wayto faint objects science and first results on extragalactic objects have made it areality. The first decade of XXI century is also remarkable for aperture synthesis inthe visual and near-infrared wavelength regimes, which provided image reconstruc-tions from stellar surfaces to Active Galactic Nuclei. Here I review the numerousastrophysical results obtained up to date, except for binary and multiple stars milli-arcsecond astrometry, which should be a subject of an independent detailed review,taking into account its importance and expected results at micro-arcsecond precisionlevel. To the results obtained with currently available interferometers, I associatethe adopted instrumental settings in order to provide a guide for potential usersconcerning the appropriate instruments which can be used to obtain the desiredastrophysical information.

Key words. instrumentation: interferometers – methods: observational – tech-niques: high angular resolution

1. INTRODUCTION

The new generation of ground-based instru-ments for high angular resolution from visual andinfrared interferometry provided a qualitatively newinformation for improving our understanding of var-ious astrophysical objects through the comparison ofobservational results with the predictions of theoret-ical models. Althoug the optical interferometry hasstill some limitations in application to various prob-lems (mainly due to the lack of sufficient signal-to-

noise ratio for faint objects), it’s range of applicationshows a significant growth, particularly in the pastdecade.

In Section 2 of the present paper I review sev-eral results that represent important contribution tothe study of the Sun and Solar system, while the re-sults concerning the determination of stellar funda-mental parameters are presented in Section 3, andrelated asteroseismological results concerning NonRadially Pulsating stars are described in Section 4.The past decade has seen some astonishing resultson stellar rotation, which I present in Section 5. The

1

S. JANKOV

related Section 6 concerns Be stars and their cir-cumstellar environements while Section 7 describesachievements in the study of Red giants and Super-giants, considering the case of Betelgeuse as the mostrepresentative one. Section 8 deals with the Asymp-totic Giant Branch stars, particularly Cepheids andMira type stars. Some remarkable results obtained inthe study of Planetary nebulae and their central starsare presented in Section 9, while Section 10 describesthe results concerning Luminous Blue Variables andfocusing to η Carinae, probably the most studied starafter our Sun. The phenomenon of Novae has beenonly recently studied by means of optical interfer-ometry and this new results are presented in Section11. Section 12 is dedicated to Young Stellar Ob-jects, namely Herbig Ae/Be stars, T Tauri stars, FUOrionis stars and Massive Young Stellar Objects, allof them being very important for our understand-ing of how accretion discs evolve into protoplanetarydiscs and finally to debris discs and planets. Theresults concerning debris discs and planet formationare presented in Section 13, while Section 14 treatscontributions of optical interferometry to the verypoupular search for exoplanets. Section 15 describesthe results obtained in the study of Central regions ofour Galaxy and Section 16 presents the compilationof contributions dedicated to Active Galactic Nuclei.Since the image reconstruction became one of themost valuable tools which generally contributes toall astrophysical fields treated here, I review some ofoutstanding results in Section 17. Finally, Section18 presents the conclusion on this work based on allresults obtained up to date.

2. SUN AND SOLAR SYSTEM

The Sun and Solar system have not been theprime candidates for application of astronomical op-tical interferometry due to to the fact that the spa-tial resolution is provided by the proximity of thisscientific targets. In the past, the single apertureinterferometry has been contributing to study of So-lar system and Sun; for example, the solar featureswith sizes of the order of 100 km or smaller werefound by means of speckle imaging (Harvey 1972,Harvey and Breckinridge 1973). From observationsof photospheric granulation from the disc center tothe limb, Wilken et al. (1997) found a decreaseof the relative contrast of the center-to-limb gran-ular intensity. A time series of high spatial resolu-tion images revealed the highly dynamical evolutionof the sunspot fine structure, namely, umbral dots,penumbral filaments, facular points (Denker 1998).Small-scale brightening in the vicinity of sunspots,was also observed in the wings of strong chromo-spheric absorption lines. These structures which areconcomitant with strong magnetic fields show bright-ness variations close to the diffraction-limit (∼0.16arcsec at λ = 550nm), of the Vacuum Tower Tele-scope, Observatorio del Teide (Tenerife). With thespeckle method, Seldin et al. (1996) found the pho-

tosphere to be highly dynamic at scales below 0.3arcsec. Speckle imaging has been successful in re-solving the Pluto-Charon system (Bonneau and Foy1980), as well as in determining shapes of asteroids(Drummond et al. 1988).

However, the investigation of Sun and So-lar system by the means of optical interferometryhas stalled, mainly due to the fact that satelliteobservatories and Solar system probes (which arenot limited by atmospheric turbulence) provided avery high spatial resolution. But, the application ofinterferometry is not excluded as shown by Delboet al. (2009) who obtained the first successful in-terferometric measurements of asteroid sizes andshapes by means of the Very Large Telescope In-terferometer (VLTI) at European Southern Observa-tory (ESO), and it’s MID-infrared Interferometric in-strument (MIDI). They observed, as a typical benchmark, the asteroid (951) Gaspra, visited in the pastby the Galileo space probe, and they derived a shapein good agreement with the results coming from thein situ measurements by the Galileo mission (see Fig.1). They have also observed the asteroid (234) Bar-bara, known to exhibit unusual polarimetric proper-ties, and they found evidence of a potential binarynature. In particular, their data were best fitted bya system of two bodies of 37 and 21 km in diameter,separated by a center-to-center distance of ∼ 24 km(projected along the direction of the baseline at theepoch of their observations).

Fig. 1. Interferometricaly deduced shape and imageof (951) Gaspra taken by the Galileo mission. Com-parison of an image of (951) Gaspra taken by theGalileo mission on 29/10/1991 at 22:26 UT from adistance of 5300 km and the shape model of Delboet al. 2009 (dashed line) observed under the samecircumstances.

2

OPTICAL INTERFEROMETRY II

3. STELLAR FUNDAMENTALPARAMETERS

Interferometry has long been a valuable toolin the study of stellar physics, primarily in obtain-ing a broad sample of measurements of fundamentalstellar parameters: radius, temperature, luminosity,mass, and chemical composition. However, they arestill not well known for a broad range of stellar types.Some properties can be measured by traditional as-tronomical means. For example, the chemical com-position at the surface can be determined by study-ing spectral lines. A star’s temperature can be de-termined from its luminosity and physical size. Todetermine the size and the mass, astronomers requireextremely high-resolution data that are not easily ob-tained except by interferometry.

A compilation of direct measurements by highangular resolution methods, as well as indirect es-timates of stellar diameters can be found in theCatalog of High Angular Resolution Measurements(CHARM, Richichi and Percheron 2002) which in-cludes most of the measurements obtained by thetechniques of lunar occultations and long-baseline in-terferometry at visual and infrared wavelengths, andwhich have appeared in the literature or were other-wise made public until mid-2001.

Star’s mass is a crucial parameter in stellar as-trophysics, however this valuable information cannotbe easily obtained directly for a single star. Interfer-ometers bring a new level of resolution to bear onspectroscopic binaries, enabling the full extractionof physical parameters for the component stars withhigh accuracy. In the case of double-lined systems,a geometrically determined orbital parallax becomesavailable as presented in the CHARA (Center forHigh Angular Resolution Astronomy) Catalog of Or-bital Elements of Spectroscopic Binary Stars (Tayloret al. 2003).

Using the Mark III Stellar Interferometer onMount Wilson, Mozurkewich et al. (2003) obtainedobservations of 85 stars at wavelengths between 451and 800 nm. Angular diameters were determinedby fitting a uniform-disc model to the visibility am-plitude versus projected baseline length. Half theangular diameters determined at 800 nm have for-mal errors smaller than 1 %. Further, the VLTIinterferometer and its VINCI and AMBER near-infrared recombiners were used, together with lit-erature measurements, to examine the luminosity-radius and mass-radius relations for K and M dwarfs,Demory et al. (2009) obtained the precision of inter-ferometric radii which competes with what can beobtained for double-lined eclipsing binaries.

Following this work, an updated Cata-log of High Angular Resolution Measurements(CHARM2), which includes results available untilJuly 2004, has been presented by Richichi et al.(2005) and complemented by the catalog of the an-gular sizes of dwarf stars and subgiants (Kervella etal. 2008) who compiled the existing long-baseline in-terferometric observations of nearby dwarf, and sub-giant stars and the corresponding broadband pho-tometry in the Johnson and Cousins bands. For the

Hyades giants, γ , δ1, ², and θ1Tau, from interfero-metric measurements with the CHARA Array, Boya-jian et al. (2009) deduced the limb-darkened angulardiameters with errors less than 2 %. In combina-tion with additional observable quantities, they de-termined the effective temperatures, linear radii, andabsolute luminosities for each of these stars, provid-ing a new calibration of effective temperatures witherrors well under 100 K.

At this precision level, the study of stellar at-mospheres became possible as shown by Wittkowskiet al. (2004a) who presented K-band interferometricmeasurements of the limb-darkened intensity profileof the M4 giant star ψ Phoenicis obtained with theVLTI/ VINCI instrument. High-precision squaredvisibility amplitudes in the second lobe of the visibil-ity function were obtained employing two 8.2 m UnitTelescopes. In addition, they sampled the visibilityfunction at small spatial frequencies using the 40cmtest siderostats (see Fig. 2). Their measurement con-strained the diameter of the star and its center-to-limb intensity variation, as well as (together with theHipparcos parallax and evolutionary models) stellarmass and surface gravity.

Fig. 2. ψ Phe diameter and limb darkening de-termination. The left panel shows the full range ofthe visibility function while the right panel is an en-largement of the low squared visibility amplitudes inthe second lobe. Measured squared visibility ampli-tudes of ψ Phe are shown (symbols with error bars)together with the (solid black line) spherical radia-tive transfer model prediction with model parametersTeff log g and mass as derived from spectrophotometryand model evolutionary tracks, and best fitting angu-lar diameter value. Shown are also the (dotted anddashed-dotted line) two plane-parallel radiative trans-fer models, all with corresponding model parametersand best fitted angular diameter. As a reference forthe strength of the limb-darkening, the gray lines de-note the corresponding uniform disc (upper line) andfull darkened disc (lower line) model visibility func-tions.

Van Belle et al. (2007) performed the directangular size measurements of the G0 IV subgiant ηBoo from the Palomar Testbed Interferometer (PTI),deducing a limb-darkened angular size of a bolo-metric flux which provided an effective temperatureand luminosity for this object. In conjunction withthe mass estimate from the MOST asteroseismologysatellite investigation, a surface gravity is establishedfor this star. Using the Infrared Optical TelescopeArray (IOTA) and it’s IONIC3 recombiner, Lacour

3

S. JANKOV

et al. (2008) performed spectro-interferometric ob-servations (R = λ/∆λ ≈ 35) of Arcturus. Imagereconstruction was performed using two software al-gorithms: Wisard and Mira (e.g. Thiebaut and Gio-vannelli 2010). No companion was detected from theclosure phases with an upper limit on the brightnessratio of 8x10−4 at 1 AU. The upper limit was also de-rived for the level of brightness asymmetries presentin the photosphere.

The high precision of direct measurements ofstellar angular diameter by the VINCI instrumenton VLTI stimulated studies which allow the applica-tions in the asteroseismology: Kervella et al. (2003a)compared the first direct angular diameter measure-ments obtained on α Centauri to recent model diam-eters constrained by asteroseismic observations. Theangular diameters of the two main components of thesystem (α Cen A and B), were measured with a rel-ative precision of 0.2% and 0.6% respectively whileBigot et al. (2006) improved the angular diameter ofα Cen B using limb-darkening predictions from a 3Dhydrodynamical radiative transfer model of its atmo-sphere to fit the interferometric visibility measure-ments. Pijpers et al. (2003) have studied the radiusof the nearby star τ Cet and the limb-darkened discdiameter is determined with an internal precision of0.5%. Kervella et al. (2003b) reported a direct mea-surement of the angular diameter of the bright starSirius A. They obtain a uniform disc angular diame-ter and (in combination with the Hipparcos parallax)a linear diameter. Using the published properties ofSirius A, they derived internal structure models cor-responding to ages between 200 and 250 ± 12 Myr.

The prospects for using the asteroseismologyof stars were hampered by the large uncertaintyin fundamental stellar parameters. Particularly forrapidly oscillating Ap (roAp) results in the litera-ture for the effective temperature (Teff) often span arange of 1000 K. In order to reduce systematic errorsand improve the Teff calibration of Ap stars Brunttet al. (2008) performed the first detailed interfer-ometric study of a rapidly oscillating roAp star, αCir. They used the Sydney University Stellar Inter-ferometer (SUSI) to measure the angular diameterand accurate Hipparcos parallax to determine the ra-dius constraining the bolometric flux from calibratedspectra which determine an effective temperature,and thus provided the first direct determination ofthe temperature of an roAp star. Further, Brunttet al. (2010) obtained long-baseline interferometricobservations of β CrB using the CHARA/FLUORinstrument. By combining the flux of the A compo-nent with its measured angular diameter, they de-termined the effective temperature Teff(A) = 7980 ±180 K (2.3 %), with uncertainty in effective temper-ature similar to that of α Cir (± 170 K).

Observing with the CHARA/VEGA spectro-interferometer, Bigot et al. (2011) measured theangular diameter of the CoRoT satellite targetHD49933. A three-dimensional radiative hydrody-namical modeling was used to compute the limbdarkening and to derive a reliable diameter. Theother fundamental stellar parameters (mass, age, and

Teff) were found by fitting the large and small p-mode frequency separations using a stellar evolutionmodel.

4. NON RADIALLY PULSATING STARS

The important observational constraints arenecessary for detailed studies of the atmosphericstructure and pulsation as shown by Kervella et al.(2004a) who reported the angular diameter measure-ment obtained with the VLTI/VINCI instrument onthe nearby star Procyon A (α CMi A, F5IV-V), at arelative precision of ± 0.9 %. They used deduced lin-ear diameter (with the Hipparcos parallax) in com-bination with the spectroscopic, photometric and as-teroseismic data to constrain the model of this star:age, an initial helium mass fraction Yi, and an initialmass ratio of heavy elements to the hydrogen. Theyalso computed the adiabatic oscillation spectrum ofProcyon A giving a mean large frequency separationin agreement with the seismic observations by Martićet al. (2004). The interferometric diameter and theasteroseismic large frequency spacing together sug-gest a mass closer to 1.4 M¯ rather than to 1.5 M¯.From this analysis, they conclude that Procyon iscurrently ending its life on the main sequence, as itsluminosity class indicates. Thévenin et al. (2005)used VLTI/VINCI angular diameter measurementsand constrained the evolutionary status of three as-teroseismic targets: δ Eri, ξ Hya, η Boo. The mainstellar modeling parameters: mass, age and metallic-ity were adjusted to fit observational data (effectivetemperature Teff , luminosity L and surface metallic-ity [Z/X]surface). In Fig. 3 representing the zoom

Fig. 3. δ Eri evolutionary status and asteroseis-mology. Zoom of the evolutionary tracks in the H-Rdiagram for δEri from label ”a” (6140 Myr) to label”j” (6230 Myr), shown by lower case letters and tri-angles with time steps of 10 Myr (except label ”G” at6200 Myr labeled by an upper case letter). The rect-angular error boxes are derived from the values andaccuracies of the adopted stellar parameters while themeasured radius and its confidence interval appear asdiagonal lines. The best model is close to label ”f” at6194 Myr. The mean large frequency splitting foundfor the best model is 45.27 µHz.

4

OPTICAL INTERFEROMETRY II

of evolutionary tracks on HR diagram for δ Eri, therectangular error boxes are derived from the val-ues and accuracies of the adopted stellar parameterswhile the measured radius and its confidence inter-val (appearing as diagonal lines) allowed the accuratedetermination of stellar evolutionary status.

Measuring the visual orbit with the Mark IIIoptical interferometer and the Navy Prototype Op-tical Interferometer (NPOI), combining it with theHipparcos proper-motion-based parallax and deter-mining the masses and magnitude difference of thecomponents of the Hyades spectroscopic binary θ2Tauri, Armstrong et al. (2006) found that both com-ponents appear to be less massive and/or brighterthan predicted from some recent evolutionary mod-els. The small scale of the λ Vir orbit (∼ 20 mas)was resolved by Zhao et al. (2007) using the IOTAinterferometer which, together with spectroscopicdata, allowed to determine its elements, as well asthe physical properties of the components. The ac-curately determined properties (0.7% and 1.5% er-rors, for A and B component respectively) allowedcomparisons between observations and current stel-lar evolution models, and reasonable matches arefound. The best-fit stellar model gives λ Vir a sub-solar metallicity of Z=0.0097 and an age of 935 Myr.The orbital and physical parameters of the star alsoallowed to study its tidal evolution timescales andstatus.

North et al. (2007) used the Sydney Univer-sity Stellar Interferometer (SUSI) to measure the an-gular diameter of β Hydri. This star is a nearbyG2 subgiant whose mean density was measured withhigh precision using asteroseismology. They deter-mined the radius and effective temperature of thestar and, by combining the radius with the meandensity, made a direct estimate of the stellar massfinding a value of 1.07 ± 0.03M¯(2.8 %). Thisvalue agrees with published estimates based on fit-ting in the Hertzsprung-Russell diagram but hasmuch higher precision, which places valuable con-straints on theoretical models of β Hyi and its oscil-lation frequencies. Mazumdar et al. (2009) studiedthe GIII red giant star ² Oph which has been foundto exhibit several modes of oscillation by the MOSTsatellite. They interpret the observed frequencies ofoscillation in terms of theoretical radial p-mode fre-quencies of stellar models, and evolutionary modelsof this star, in both the shell H-burning and core He-burning phases of evolution, are constructed. Theyalso obtained an independent estimate of the photo-spheric radius with highly accurate interferometricobservations in the infrared K’ band (1.9-2.3 µm)by using the CHARA/FLUOR instrument togetherwith the Hipparcos parallax. The radius obtainedfrom the asteroseismic analysis matched the inter-ferometric value quite closely even though the radiuswas not constrained during the modeling.

However, with classical interferometric arraysit is possible to resolve spatially only a limited num-ber of nearby bright stars and even more difficultto measure the radii with sufficient precision for as-teroseismology. On the other hand, the high resolu-tion spectroscopy allows for (through Doppler Map-ping) indirect observational information on atmo-

spheric structures of an unresolved star by model-ing the observed flux distribution across the spec-tral lines. For instance, the differential interferome-try (Beckers 1982, Petrov 1988) makes it possible tomeasure the shift of the stellar photometric barycen-ter (photocenter) of an unresolved star as a functionof wavelength, providing the first order moment ofthe spatial brightness distribution in addition to thezero order moment spectroscopic information, andallowing better spatial resolution of non-radial stel-lar pulsations when compared to the classical inter-ferometric and Doppler imaging methods (Jankov etal. 2001, 2002).

Despite of a remarkable progress in under-standing stellar interiors, we know surprisingly littleabout the internal structure of stars spinning neartheir critical limit. New interferometric imaging ofthese rapid rotators (see Section 5.) combined withadvances in asteroseismology promises to lift thisveil and probe the strongly latitude-dependent pho-tospheric characteristics and even reveal the internalangular momentum distribution of these objects. Us-ing the physical properties of α Oph, recently deter-mined from long-baseline interferometry data fromthe CHARA Array, and the high-precision photome-try (based on 30 continuous days of monitoring usingthe MOST satellite) of this rapidly rotating δ Scutivariable star, Monnier et al. (2010) have identified57 ± 1 distinct pulsation modes. Remarkably, theyhave also discovered that the fast rotation modulatesthe low-frequency modes identified as a rich familyof g-modes (|m| up to 7). They found that the spac-ing of the g-modes is surprisingly linear consideringCoriolis forces which are expected to strongly distortthe mode spectrum suggesting prograde ”equatorialKelvin” waves (modes l = m).

5. RAPIDLY ROTATING STARS

Optical long baseline interferometry is a pow-erful tool to study detailed stellar shapes. In par-ticular, a rapid rotation induces interferometric sig-natures requiring a detailed modeling to correctlyinterpret high angular resolution data. Domicianode Souza et al. (2002) studied the effects of uniformstellar rotation on interferometric observables using aphysically coherent model that includes gravity dark-ening and geometrical deformation, as well as a ra-diation transfer code. They investigated the use ofmulti-baseline and/or multi-wavelength-channel ob-servations, both in the continuum and spectral lines,in order to obtain the unique solutions for relevantmodel parameters. They showed that this is possi-ble and provided a guide for observers to performthis task. Soon after, Domiciano de Souza et al.(2003) reported the first observations of a rapidly ro-tating Be star Achernar (α Eridani), using the Earth-rotation synthesis on the VLTI (Fig. 4). Their mea-surements correspond to an a/b = 1.56 ± 0.05 appar-ent oblate star (Fig. 5), a and b being the equivalentuniform disc angular radii in the equatorial and polardirection respectively. Considering the presence

5

S. JANKOV

Fig. 4. Achernar’s observations with VLTI/VINCI.VLTI ground baselines for Achernar observationsand their corresponding projections onto the skyat different observing times. Left: Aerial view ofVLTI ground baselines for the two pairs of 40 cmsiderostats used for Achernar observations. Right:Corresponding baseline projections onto the sky asseen from the star.

Fig. 5. Achernar’s shape reconstruction. Fit ofan ellipse over the observed squared visibilities ofAchernar translated to equivalent uniform disc an-gular diameters. Each value is plotted together withits symmetrical azimuthal value. The fitted ellipsereveals an extremely oblate shape with an equatorialto polar ratio a/b = 1.56± 0.05.

of a circumstellar envelope they argued that theirmeasurement corresponds to a truly distorted star,since α Eridani exhibited a negligible Hα emis-sion during their interferometric observations. Inthis framework, they conclude that the commonlyadopted Roche approximation (uniform rotation andcentrally condensed mass) should not apply to this

star. This result opened new perspectives to basicastrophysical problems such as the rotationally en-hanced mass loss and internal angular momentumdistribution. In addition to its intimate relation withmagnetism and pulsation, a rapid rotation thus pro-vides a key to the Be phenomenon (see Section 6):one of the outstanding non-resolved problems in stel-lar physics.

This astonishing result has been confirmedtheoretically (Jackson et al. 2004) and observation-ally: McAlister et al. (2005) reported K-band inter-ferometric observations of the bright, rapidly rotat-ing star Regulus (type B7 V) made with the CHARAinterferometer, deducing a high stellar obleteness of1.32± 0.06 (Fig. 6).

Fig. 6. K-band image of Regulus. The fitted ellipserevealed an extremely oblate shape with an equatorialto polar ratio 1.32± 0.06.

Infrared interferometric angular size measure-ments using narrowband channels in the spectrom-eter at PTI (Palomar Testbed Interferometer, MtPalomar), indicated a non-circular projected discbrightness distribution for the A7IV-V star Altair(val Belle et al. 2001). Given the known rapid ro-tation of this star, they modeled the data as arisingfrom an elongated rigid atmosphere. To the first or-der, an ellipse with an axial ratio of a/b=1.140 ±0.029 could be fitted to their interferometric diame-ter measurements. In addition, Ohishi et al. (2004)reported an asymmetric surface brightness distribu-tion of this rapidly rotating star measured by theNPOI instrument. The outstanding characteristicsof these observations were the high resolution withthe minimum fringe spacing of 1.7 mas, easily resolv-ing the 3 mas stellar disc, and the measurement ofthe closure phase, which is a sensitive indicator ofthe asymmetry of the brightness distribution of thesource. They fitted the measured observables to amodel with a bright spot on a limb-darkened disc

6

OPTICAL INTERFEROMETRY II

and found that the observations are well reproducedby a bright spot which has relative intensity of 4.7%, on the limb-darkened stellar disc. Rapid rotationof Altair indicated that this bright region is a pole,which is brighter than other part of the star owingto gravity darkening. Including new data (squaredvisibilities in the H and K bands from VLTI/VINCI)as well as previously published data (squared visibil-ities in the K band from PTI and squared visibili-ties, triple amplitudes, and closure phases in the vis-ible from NPOI), Domiciano de Souza et al. (2005)showed that the observations can only be explainedif Altair has a gravity-darkening compatible with theexpected value for hot stars, i.e., the von Zeipel effect(Teff ∼ g0.25).

Van Belle et al. (2006) presented observationsof the A7 IV-V star Alderamin (α Cep) from theCHARA Array. The infrared interferometric angu-lar size measurements indicated a non-circular pro-jected disc brightness distribution for this rapid ro-tator. The interferometric observations were mod-eled as arising from an elongated rigid atmospherewith an axial ratio of 1.298 ± 0.051. The inclinationof Alderamin to the line of sight indicated by thismodeling is (i = 88.2◦+1.8−13.3) and the star has a rota-tional velocity which is ∼ 83 % of breakup velocity.A remarkable aspect of this modeling was a deter-mination of the gravity-darkening coefficient which,at a value of β = 0.084+0.026−0.049, is consistent with aconvective photosphere as expected for an A7 IV-Vstar.

Peterson et al. (2006a) reported the success-ful fitting of a Roche model with a surface tempera-ture gradient following the von Zeipel gravity dark-ening law to observations of Altair made with theNavy Prototype Optical Interferometer. They con-firm the claim by Ohishi et al. (2004) and Domicianode Souza et al. (2005) that Altair displays an asym-metric intensity distribution due to rotation. Themodeling indicated that Altair is rotating at 0.90 ±0.02 of its breakup (angular) velocity. Their resultsare consistent with the apparent oblateness found byvan Belle et al. and show that the true oblatenessis significantly larger owing to an inclination of therotational axis of ∼ 64◦. Of particular interest, theyconclude that instead of being substantially evolvedas indicated by its classification A7 IV-V, Altair isonly barely off the zero-age main sequence and rep-resents a good example of difficulties rotation can in-troduce in the interpretation of this part of the HRdiagram.

Peterson et al. (2006b) reported a closurephase optical interferometric observations at theNPOI that show that Vega has the asymmetricbrightness distribution of the bright, slightly off-set polar axis of the star rotating at 93% of itsbreakup velocity. In addition to explaining the un-usual brightness and line shape peculiarities, thisresult leads to the prediction of an excess of near-infrared emission compared to the visible, in agree-ment with observations. The large temperature dif-ferences predicted across its surface call into ques-tion composition determinations, adding uncertaintyto Vega’s age and opening the possibility that its

debris disc could be substantially older than previ-ously thought. Yoon et al. (2008) reported a reanal-ysis of Vega’s composition. A full spectral synthesisbased on the Roche model derived earlier from NPOIinterferometry is used. They find the line shapesin Vega’s spectrum to be more complex than justflat-bottomed, which have been previously reported.They investigate the effects of rotation on the de-duced abundances and show that the dominant ion-ization states are only slightly affected compared toanalyses using nonrotating models. They argue thatthe rapid rotation requires the star to be fully mixed.This composition leads to masses and particularlyages that are quite different compared to what wereusually assumed.

Zhao et al. (2009) presented sub-milli-arcsecond resolution imaging and modeling of twonearby rapid rotators, α Cephei and α Ophiuchi,obtained with the CHARA array. Incorporating agravity-darkening model, they determine the incli-nation, the polar and equatorial radius and temper-ature, as well as the fractional rotation speed of thetwo stars with unprecedented precision. The polarand equatorial regions of the two stars have ∼ 2000K temperature gradient causing their apparent tem-peratures and luminosities to be dependent on theirviewing angles. Their modeling (Fig. 7) allowed todetermine the true effective temperatures and lumi-nosities of α Cep and α Oph permitting to investigatetheir true locations on the H-R diagram. These prop-erties in turn give estimates of the masses and ages ofthe two stars within a few percent of error using stel-lar evolution models. Also, based on their gravity-darkening modeling, they propose a new method toestimate the masses of single stars in a more directway through Ve sin i measurements and precise geo-metrical constraint.

Fig. 7. Best-fit standard gravity-darkening model ofα Oph. The contours in the left panel indicate the lo-cal brightness temperatures on the surface of the star.The right panel shows the latitude and longitude ofα Oph to help visualize its geometry.

An alternative approach to study the rapidlyrotating stars has been proposed by Jankov et al.(2003a and 2003b) who considered the case of stellaractivity, showing the potential of new methods whichcombine the classical spectroscopy (TomographicImaging) and Long Baseline Interferometry, provid-ing informations that cannot be obtained otherwiseby each of these techniques taken separately. By

7

S. JANKOV

means of numerical simulations, Rousselet-Perrautet al. (2004) investigated the ability of optical inter-ferometry via the fringe phase observable to addressstellar activity and magnetism. To derive abundancemaps and stellar rotation axes they use the colordifferential interferometry which couples high angu-lar resolution to high spectral resolution while, toconstrain magnetic field topologies, they add a po-larimetric mode. They emphasize the crucial needfor developing and validating inversion algorithms sothat future instruments on optical aperture synthesisarrays can be optimally used.

6. Be STARS

Be stars show evidence of mass loss and cir-cumstellar envelopes (CSE) from UV resonance lines,near-IR excesses, and the presence of episodic hydro-gen emission lines. The geometry of these envelopesis still uncertain, although it is often assumed thatthey are formed by a disc around the stellar equatorand a hot polar wind. This envelope emission is moreextended, and thus more easily resolvable, than thetiny photosphere itself. The envelope of γ Cas wasfirst resolved by Thom et al. (1986) using the I2Tinterferometer, and Mourard et al. (1989) using theGI2T interferometer. The high spectral resolutionof GI2T later also uncovered asymmetric emission inthese Be star envelopes in Hα (Stee et al. 1995), andobserved in other lines too (Stee et al. 1998). Vakiliet al. (1998) proposed that the emission line regionis very one-sided and time-variable.

With a good range of baselines, the Mark IIIinterferometer was able to detect definite asymme-tries in γ Cas and ζ Tau (Quirrenbach et al. 1993,1994) while Tycner et al. (2004) presented opticalinterferometric observations of ζ Tau obtained usingthe NPOI. The observations suggested a strong de-parture from circular symmetry which has been de-scribed by an elliptical model. Tycner et al. (2005)presented the long-baseline interferometric observa-tions, obtained with the NPOI, of the Hα-emittingenvelopes of the Be stars η Tau and β CMi, demon-strating a clear dependence of the net Hα emission onthe linear size of the emitting region. These resultsare consistent with an optically thick line emissionthat is directly proportional to the effective area ofthe emitting disc. Chesneau et al. (2005a) presentedthe first VLTI/MIDI observations of the Be star αAra showing a nearly unresolved circumstellar discin the N band. These measurements put an upperlimit on the envelope size corresponding to 14 R∗assuming R∗=4.8 R¯ and the Hipparcos distance of74 pc. These observations also placed complemen-tary constraints on the density and geometry of theα Ara circumstellar disc.

Tycner et al. (2006) presented interferometricobservations of two well-known Be stars, γ Cas andϕ Per, collected and analyzed to determine the spa-tial characteristics of their circumstellar regions. Theobservations were obtained using the NPOI equippedwith narrowband filters which isolate the Hα emis-sion line from the nearby continuum radiation, re-sulting in an increased contrast between the inter-

ferometric signature due to the Hα-emitting circum-stellar region and the central star, thus allowing theinterferometric signal in the Hα channel to be cali-brated with respect to the continuum channels. Theobservations used in this study represent the highestspatial resolution measurements of the Hα-emittingregions of Be stars obtained to date. These observa-tions allowed them to demonstrate for the first timethat the intensity distribution in the circumstellarregion of a Be star cannot be represented by a uni-form disc or ring-like structures whereas a Gaussianintensity distribution appeared to be fully consistentwith their observations. Kervella and Domiciano deSouza (2006a) presented long-baseline interferomet-ric observations of Achernar with the VLTI/VINCIbeam recombiner in the H and K bands by usingvarious telescope configurations and baseline lengthswith a wide azimuthal coverage. They clearly de-tected a CSE elongated along the polar axis of thestar as well as rotational flattening of the stellar pho-tosphere. They conclude that this CSE could belinked to free-free emission from the radiative pres-sure driven wind originating from the hot polar capsof the star.

Gies et al. (2007) presented the first K’-band,long-baseline interferometric observations of the Bestars γ Cas, ϕ Per, ζ Tau, and κ Dra. The measure-ments were made with multiple telescope pairs ofthe CHARA interferometer resolving the circumstel-lar discs of the targets. The disc resulting densitiesare in broad agreement with prior studies of the IRexcess flux, and the resulting orientations generallyagree with those from interferometric Hα and contin-uum polarimetric observations. They find that theangular size of the K’ disc emission is smaller thanthat determined for the Hα emission, and they arguethat the difference is the result of a larger Hα opac-ity and the relatively larger neutral hydrogen frac-tion with increasing disc radius. Domiciano de Souzaet al. (2007) presented the first high spatial andspectral resolution observations of the circumstellarenvelope of a B[e] supergiant (CPD-57◦2874), per-formed with the VLTI. Spectra, visibilities and clo-sure phase were obtained using the beam-recombinerinstruments AMBER, near-IR interferometry withthree 8.3 m Unit Telescopes (UTs) and MIDI withtwo UTs. The interferometric observations of theCSE are well fitted by an elliptical Gaussian modelwith FWHM diameters varying linearly with wave-length. The major-axis position angle of the elon-gated CSE in the mid-IR agrees well with previouspolarimetric data, hinting that the hot-dust emissionoriginates in a disc-like structure and supporting thenon-spherical CSE paradigm for B[e] supergiants.

Using the VLTI/AMBER instrument operat-ing in the K-band, Meilland et al. (2007a) studiedthe geometry and kinematics of the disc around theBe star α Arae as a function of wavelength, espe-cially across the Brγ (λ 2.1657 µm) emission line,which provided a gain by a factor of 5 in spatial res-olution compared to previous VLTI/MIDI observa-tions. Consequently, it was possible to combine thehigh angular resolution provided with the medium(R∼ 1500) spectral resolution of AMBER to studythe kinematics of the inner part of the disc and to in-

8

OPTICAL INTERFEROMETRY II

fer its rotation law. For the first time, they obtaineddirect evidence that the disc is in the Keplerian ro-tation, answering the question that has existed sincethe discovery of the first Be star (γ Cas). They foundthat the disc around α Arae is compatible with adense equatorial matter confined to the central re-gion whereas a polar wind is contributing along therotational axis of the central star (Fig. 8).

Fig. 8. Models of α Arae. The four upper panelsare a cut of the circumstellar disc in a plane definedby the observer line of sight and the stellar rotationalaxis (the observer is on the right in each picture);the corresponding projections onto the sky-plane withthe interferometric data points from MIDI and AM-BER over-plotted are the central pictures, whereas a3Dview’ is plotted into the lower row for each model.

Meilland et al. (2007b) observed withVLTI/AMBER the circumstellar environment of theBe star κ CMa in the Brγ emission line and its nearbycontinuum to study the kinematics within the discand to infer its rotation law. Using differential vis-ibilities and differential phases across the Brγ linethey detected an asymmetry in the disc (see Fig. 9).Moreover, they found that κ CMa seems difficult tofit within the classical scenario for Be stars, illus-trated by previous α Arae observations, i.e. a fast ro-tating B star close to its breakup velocity surroundedby a Keplerian circumstellar disc with an enhancedpolar wind. They discuss the possibility that κ CMais a critical rotator with a Keplerian rotating discand examine whether the detected asymmetry canbe interpreted within the ”one-armed” viscous discframework.

Meilland et al. (2008) studied the Be star δCen circumstellar disc in the H and K band usinglow (R=35) and medium (R=1500) spectral resolu-tion observations. They detected an oscillation in thevisibility curve plotted as a function of the spatialfrequency which is a clear signature of a companionaround the star. They also report an envelope fluxaround the Be primary that contributes up to about50 % of the total flux, in agreement with SpectralEnergy Distribution (SED). The envelope size wasestimated but no departure from spherical symme-try was detected.

Fig. 9. κ CMa intensity map in the contin-uum at 2.15µm. The inclination angle is 60◦, thecentral black dot represents the κ CMa photosphere(0.25mas).

Using the VLTI/VINCI instrument, Carciofiand Domiciano de Souza (2008) performed a new in-terferometric study of Achernar, identifying two dif-ferent disc models that simultaneously fit the spec-troscopic, polarimetric, and interferometric observa-tional constraints: a tenuous disc in hydrostatic equi-librium (i.e. with small scale height) and a smaller,scale height enhanced disc. They concluded thatthese relatively small discs correspond to the tran-sition region between the photosphere and the cir-cumstellar environment and that they are probablyperturbed by some photospheric mechanism.

Kervella et al. (2009) searched for the signa-ture of circumstellar emission at distances of a fewstellar radii from Achernar, in the thermal IR do-main. They obtained interferometric observationson three VLTI baselines in the N band (8-13 µm),using the MIDI instrument. From the measured vis-ibilities, they derive the angular extension and fluxcontribution of the N band circumstellar emission inthe polar direction of Achernar. This flux contribu-tion is in good agreement with the photometric IRexcess measured by fitting the spectral energy dis-tribution. They concluded that this polar envelope,already detected at 2.2 µm, is most probably an ob-servational signature of the fast wind ejected by thehot polar caps of the star.

Meilland et al. (2009) obtained calibratedvisibility measurements for stars: p Car, ζ Tau,κ CMa, α Col, δ Cen and β CMi, α Ara, usingthe VLTI/MIDI instrument in the N band. Theycompared their results with previous K band mea-surements obtained with the VLTI/AMBER instru-ment and/or the CHARA interferometer, concludingthat the size of the circumstellar envelopes for theseclassical Be stars does not seem to vary stronglyon the observed wavelength between 8 and 12 µm.Millan-Gabet et al. (2010) presented near-infraredH and K-band spectro-interferometric observationsof the gaseous disc around the primary Be star inthe δ Sco binary system. Using observations at theCHARA/MIRC instrument in the H band they re-solved an elongated disc, while using the Keck Inter-

9

S. JANKOV

ferometer (KI) the source of the K-band continuumemission was only marginally spatially resolved. Onthe other hand, the line emission in He I λ 2.0583µm and Brγ, was clearly detected with ∼ 10 % lowervisibilities than those of the continuum. When tak-ing into account the continuum/line flux ratio thistranslates into much larger sizes for the line emis-sion regions. Their KI data also reveal a relativelyflat spectral differential phase response, ruling out asignificant off-center emission.

Meilland et al. (accepted 2011) carried outobservations of the sample composed of eight brightclassical Be stars: α Col, κ CMa, ω Car, pCar, δ Cen, µ Cen, α Ara, and o Aqr, with theVLTI/AMBER instrument combining high spectral(R=12000) and high spatial (θmin=4mas) resolu-tions. They determined the disc extension in theline and the nearby continuum for most targets, con-straining the disc kinematics and showing that it isdominated by rotation with a rotation law close tothe Keplerian one. The survey also suggests thatthese stars are rotating below their critical velocities(Vc) with a mean rotational rate of 0.82±0.08 Vc,leading to the conclusion that their Be stars samplesuggests that the rotation alone cannot explain theorigin of the Be phenomenon and that other mecha-nisms are playing a non-negligible role in the ejectionof matter.

7. RED GIANTS ANDSUPERGIANTS: Betelgeuse

Historically, red giants and supergiants wereone of the favorite targets for stellar interferometry,and there are many of them for which the funda-mental parameters are precisely derived. However,the most famous case in the quest for surface inho-mogenities in giants stars is the M2 supergiant Betel-geuse (α Ori). The interferometry research to studythe inhomogenities of stellar surfaces began by us-ing (visible-light) aperture masking on the WilliamHerschel telescope in the Canary Islands (Baldwinet al. 1986, Haniff et al. 1987). Interferometricnon-redundant masking imaging was performed andearly results showed bright features (strong depar-ture from circular symmetry) on the surface of theBetelgeuse (Buscher et al. 1990), confirming someprevious reports (e.g. Roddier and Roddier 1983).No long baseline (separate-element) interferometerwould be able to investigate the nature of these fea-tures for years, and the Cambridge masking grouphas spent more than a decade thoroughly investigat-ing ”hotspots” on red supergiants and giants. Overthe two last decades, it was shown that asymmetriesare common (although not omnipresent) around redsupergiants and giants at visible wavelengths (Wil-son et al. 1992, Tuthill et al. 1997, 1999), thatthese hotspots vary on a timescale of months (Wil-son et al. 1997), and that the asymmetries be-come less-pronounced (even disappearing) into theIR (Young et al 2000). However, the first image ofa stellar photosphere using optical aperture synthe-

sis with the Cambridge Optical Aperture SynthesisTelescope (COAST) interferometer showed a feature-less (uniform and circularly symmetric disc) Betel-geuse (Burns et al. 1997).

Ohnaka et al. (2009) presented spatially re-solved, high-spectral resolution K-band observationsof the Betelgeuse using VLTI/AMBER in order toprobe inhomogeneous structures in the dynamicalatmosphere of the star. Betelgeuse was observed inthe wavelength range between 2.28 and 2.31 µm withspectral resolutions of 4800-12000 allowing to studyinhomogeneities seen in the individual CO first over-tone lines. The data could roughly be explained by asimple model in which a patch of CO gas is movingoutward or inward with velocities of 10-15 km s−1,while the CO gas in the remaining region in the at-mosphere is moving in the opposite direction at thesame velocities. Also, the data are consistent withthe presence of warm molecular layers extending to1.4-1.5 R∗. Haubois et al. (2009) reported on H-bandinterferometric observations of Betelgeuse made bythe three-telescope interferometer IOTA. They im-aged the star and its asymmetries to deduce the spa-tial variation of the photosphere, including its di-ameter, limb darkening, effective temperature, sur-rounding brightness, and star spots. Resolved im-ages of Betelgeuse in the H band are asymmetricat the level of a few percent while the amount ofmeasured limb-darkening is in good agreement withmodel predictions. The two spots imaged at the sur-face of the star are potential signatures of convectivecells (Fig. 10).

Fig. 10. Image reconstruction of Betelgeuse. Upperpanel: contour image reconstruction of Betelgeusefrom MIRA. Lower panel: contour image reconstruc-tion from WISARD of Betelgeuse in a 60 mas field.Both images were reconstructed with the same apri-ori object and the same type of regularization.

10

OPTICAL INTERFEROMETRY II

Ravi et al. (2011) presented an interferomet-ric study of the continuum surface of Betelgeuseat 11µm wavelength, using data obtained with theBerkeley Infrared Spatial Interferometer (ISI) eachyear between 2006 and 2010. These data allowed aninvestigation of an optically thick layer within 1.4R∗, which has an optical depth of ∼ 1 at 11 µm,and varies in temperature between 1900 K and 2800K and in outer radius between 1.16 and 1.36 stellarradii. The layer has a non-uniform intensity distribu-tion that changes between observing epochs. Theseresults indicate that large-scale surface convective ac-tivity strongly influences the dynamics of the inneratmosphere of Betelgeuse and mass-loss processes.

8. ASYMPTOTIC GIANTBRANCH STARS

The apparent interferometric sizes of variablered giants vary dramatically up to a factor of 3 withwavelength and pulsation phase (Tuthill et al. 1995,Weiner et al. 2000, Mennesson et al. 2002, Perrin etal. 2004a, Weiner 2004). These large variations chal-lenge the hydrodynamical and line opacity modelsof these stars indicating that their extended atmo-spheres are extremely complex. This recently stim-ulated a large sample of stars to be observed usingbenefits of high spatial resolution provided by newgeneration of interferometers. Infrared interferome-try of Asymptotic Giant Branch (AGB) stars has re-cently been reinterpreted as revealing the presence ofdeep molecular layers. Empirical models for a photo-sphere surrounded by a simple molecular layer or en-velope have led to a consistent interpretation of pre-viously inconsistent data. The stellar photospheresare found to be smaller than previously understoodand the molecular layer is much higher and denserthan predicted by the hydrostatic equilibrium.

The size of the continuum photospheres of αOri, α Her, R Leo, and χ Cyg have been measuredat 11 µm by Weiner et al. (2003), using heterodyneinterferometry (obtained from ISI) to accuracies ashigh as 1%. The resulting apparent diameters forα Ori and α Her are ∼ 30 % larger than measurednear-infrared sizes, whereas the Mira variables R Leoand χ Cyg, have 11 µm apparent diameters, roughlytwice their reported near-infrared sizes. Mennessonet al. (2005) reported interferometric observationsof the semiregular variable star RS CrB, a red gi-ant with strong silicate emission features. The datawere among the first long baseline mid-infrared stel-lar fringes obtained between the Keck telescopes, us-ing parts of the new nulling beam recombiner. Thelight was dispersed by a low-resolution spectrome-ter allowing simultaneous measurement of the sourcevisibility and intensity spectra from 8 to 12 µm, andthe observations provided a non-ambiguous deter-mination of the dust shell spatial scale and relativeflux contribution. Ragland et al. (2006) have mea-sured non-zero closure phases for the sample of 56nearby AGB stars using the three-telescope IOTA in-terferometer at near-infrared wavelengths (H band)and with angular resolutions in the range 5-10 mas.Reporting the measured stellar angular sizes, they

hypothesize that most Mira stars would show de-tectable asymmetry if observed with adequate an-gular resolution since the detected non-zero closurephases can only be generated by asymmetric bright-ness distributions of the target stars or their sur-roundings. Deroo et al. (2006) presented the firstmid-IR long baseline VLTI/MIDI interferometric ob-servations of the circumstellar matter around binarypost-AGB stars, SX Cen and HD 52961, confirmingthe disc interpretation of the spectral energy distri-bution of both stars.

8.1. Cepheids

Although phenomenologically related to mea-surements of pulsating AGB stars, observations ofCepheids are quite distinct in their scientific goals.As has been discussed (e.g. Davis 1976), optical in-terferometry should play an important role in inde-pendent calibration of the Cepheid distance scale,a crucial element of the cosmic distance ladder.By measuring the changing diameter of a nearbyCepheid and the radial velocity curve through a pul-sation cycle, the distance can be directly inferred viathe Baade-Wesselink method, and unbiased angulardiameter measurements are required for accurate dis-tances to Cepheids.

The initial results have appeared from GI2T(Mourard et al. 1997), NPOI (Armstrong et al.2001), IOTA (Kervella et al. 2001), and PTI (thefirst direct detection of Cepheid pulsation; Lane etal. 2000). The field is rapidly developing, both ob-servationally and theoretically. Lane et al. (2002)presented observations of the Galactic Cepheids ηAql and ζ Gem, and their observations were able toresolve the diameter changes associated with pulsa-tion. This allows the distance to the Cepheids tobe determined independently of photometric obser-vations, providing calibration of surface brightnessrelations for use in extragalactic distance determi-nation. They also provided a measurement of themean diameter of these Cepheids which is crucialfor constructing reliable structural models of thisstellar class. Nordgren et al. (2002) used directdiameter observations of Cepheid variables to cali-brate the Barnes-Evans Cepheid surface brightnessrelation. Fifty-nine separate Cepheid diameter mea-surements from four different optical interferometerswere used to calculate surface brightnesses as a func-tion of magnitude and color. The linear relation toCepheid surface brightness versus color was in ex-cellent agreement with functions found using inter-ferometric observations of nonvariable giant and su-pergiant stars. Using these relations, they deduceddistance of δ Cephei, and compared it to an inde-pendent distance which is known from trigonometricparallax.

Kervella et al. (2004b) reported the angulardiameter measurements of seven classical Cepheids,X Sgr, η Aql, W Sgr, ζ Gem, β Dor, Y Oph andl Car, obtained with the VLTI/VINCI instrument.They also used reprocessed archive data obtainedwith the IOTA/FLUOR instrument on ζ Gem, inorder to improve the phase coverage of their obser-vations. They obtained average limb darkened angu-

11

S. JANKOV

lar diameters for four of these stars, η Aql, W Sgr,β Dor, and l Car, and they detected the pulsationalvariation of their angular diameters. This allowedto compute directly their distances, using a modifiedversion of the Baade-Wesselink method. Kervella etal. (2004c) derived new calibrations of the Cepheidperiod-radius (P-R) and period-luminosity (P-L) re-lations using the interferometric angular diametermeasurements of seven classical Cepheids reportedin Kervella et al. (2004b), and complemented bypreviously existing measurements. Taking advan-tage of the significantly larger color range covered bythese observations, they derived high precision cali-brations of the surface brightness-color relations us-ing exclusively Cepheid observations, making it pos-sible to improve the distance to Cepheids througha Baade-Wesselink type technique. However, usingthis method, the distance is known to a multiplica-tive factor called the projection factor. This factorhas been measured directly, for the first time usinginterferometry at the CHARA Array by Mérand etal. (2005) who deduced a geometrical measurementof the projection factor of δ Cep.

VLTI observations of the brightest and an-gularly largest classical Cepheid, l Carinae have re-solved with high precision the variation of its angulardiameter with phase. Kervella et al. (2004d) com-pared the measured angular diameter curve to theone that they derive by an application of the Baade-Wesselink-type infrared surface brightness techniqueand found an almost perfect agreement between thetwo curves. The mean angular diameters of l Carfrom the two techniques agree very well within theirtotal error bars (1.5 %) as do the derived distances(4 %) indicating that the calibration of the surfacebrightness relations used in the distance determina-tion of far-away Cepheids is not affected by large bi-ases. Kervella et al. (2006b) presented the results oflong-baseline interferometric observations of l Cari-nae in the infrared N (8-13 µm) and K (2.0-2.4 µm)bands, using the MIDI and VINCI instruments of theVLT Interferometer. In the N band they resolved alarge circumstellar envelope. The signature of thisenvelope was also detected in the K band data asa deviation from a single limb darkened disc visibil-ity function. Considering a possibility that this CSEis linked to the relatively large mass loss rate of lCar, they discuss an analogy with the molecular en-velopes of Red supergiants, and Miras. Mérand etal. (2006) presented the results of long-baseline in-terferometric observations of the classical CepheidsPolaris (α UMi) and δ Cep in the near infrared K’band, using the CHARA/FLOUR Array. Followingtheir previous detection of a circumstellar envelopearound l Car (Kervella et al. 2006b), they reportsimilar detections around Polaris and δ Cep.

Mérand et al. (2007) observed a Cepheid YOph for which the pulsation has been well resolvedusing the long-baseline near-infrared interferometrywith CHARA/FLOUR. They found that the obser-vations suggest the star surrounded by a circumstel-lar envelope with characteristics similar to that foundpreviously around other Cepheids, pointing towardthe conclusion that most Cepheids are surrounded

by faint circumstellar envelopes. Observations of thesouthern Cepheid l Car to yield the mean angular di-ameter and angular pulsation amplitude, have beenmade with the SUSI array at a wavelength of 696nm by Davis et al. (2009). The interferometric re-sults have been combined with radial displacementsof the stellar atmosphere derived from selected radialvelocity data to determine the distance and mean di-ameter of l Car, showing excellent agreement withpreviously published values. However, no evidencewas found for a circumstellar envelope at 696 nm.

8.2. Miras

Mira variables are low- to intermediate-massasymptotic giant branch (AGB) stars that pulsatewith a period of about one year. They have a cool(Teff≤ 3000 K) and extended (R > 100R∗) photo-sphere. The circumstellar environment of Mira vari-able stars is characterized by cool temperatures andrelatively high densities leading to a richer chemistrythan that found in hotter stars and to formation ofsolid-state dust grains.

The long-term monitoring of Mira variablesat the PTI performed by Thompson et al. (2002a)provided high-resolution narrowband angular sizes ofthe oxygen-rich Mira S Lac and the carbon-rich MiraRZ Peg in the near-infrared. Their data set spannedthree pulsation cycles for S Lac and two pulsation cy-cles for RZ Peg and represents the first study to cor-relate multi-epoch narrowband interferometric dataof Mira variables. As a part of a long-term observa-tional program using the IOTA Array, van Belle etal. (2002) presented angular size measurements of 22oxygen-rich Mira variable stars in order to character-ize the observable behavior of these stars. A simpleexamination of the resultant sizes of these stars inthe context of pulsation mode was consistent with atleast some of these objects pulsating in the funda-mental mode.

Thompson et al. (2002b) reported high-resolution (< 0.05 mas) angular size measurementsof the Mira variable star R Tri using the PTI in the Kband (2.0-2.4 µm), and modeling with simple geome-tries. For the axially symmetric models, the positionangles are roughly perpendicular to visual polariza-tion position angles, which supports an axially sym-metric source of light scattering. For the elliptical ge-ometry, the axial ratio of 1.33 has been determined,similar to that previously determined for other Miraand semiregular variable stars.

Mennesson et al. (2002) observed nine brightO-rich Mira stars and five semiregular variable coolM giants with the IOTA interferometer in both K’ (∼2.15 µm) and L’ (∼ 3.8 µm) broadband filters. Toexplain large apparent diameter increases betweenthe K’ and L’ bands they propose a simple two-component model consisting of a warm (1500-2000K), extended (up to ∼ 3 stellar radii), optically thin(τ ∼ 0.5) layer located above the classical photo-sphere which could explain the observed variationof Mira uniform disc diameters. This interpretationwas consistent with the extended molecular gas lay-ers (H2O , CO, etc.) inferred around some of theseobjects from previous IOTA K’-band interferometric

12

OPTICAL INTERFEROMETRY II

observations. The two-component model had imme-diate implications: the Mira photosphere diametersare smaller than previously recognized which implieshigher effective temperatures, favoring fundamentalmode pulsation. Using the same instrument, Hof-mann et al. (2002) presented K’-band observationsof five Mira stars: M-type Miras: X Oph, R Aql, RUHer, R Ser, and the C-type Mira V CrB. The derivedangular Rosseland radii and the bolometric fluxesallowed the determination of effective temperatureswhich, together with deduced linear radii, confirmedfundamental mode pulsation interpretation.

Perrin et al. (2004b) have observed Mira starson the IOTA interferometer in narrow bands around2.2 µm wavelength, finding systematically larger di-ameters in bands contaminated by water vapor andCO. Their visibility measurements could be inter-preted by a model comprising a photosphere sur-rounded by a thin spherical molecular layer while thededuced photospheric diameters were found smallerthan in previous studies by several tens of percent,indicating that all Mira stars are fundamental modepulsators; the previous studies leading to the conclu-sion of the first-overtone mode were biased by toolarge diameter estimates.

Woodruff et al. (2004) presented K-band ob-servations of the Mira star prototype o Cet obtainedby the VLTI/VINCI instrument and two siderostats.Their comparison of deduced Rosseland radii, effec-tive temperatures, and the shape of the observedvisibility functions with model predictions confirmedthat o Cet is a fundamental mode pulsator. Ohnakaet al. (2005) presented the results of the firstmid-infrared interferometric observations of the Miravariable RR Sco with the VLTI/MIDI, togetherwith K-band observations using VLTI/VINCI. Theirmodel calculations show that optically thick emis-sion from a warm molecular envelope consisting ofH2O and SiO can cause the apparent mid-infrareddiameter to be much larger than the continuum di-ameter. The observed increase of the uniform-discdiameter longward of 10 µm could be explained byan optically thin dust shell consisting of silicate andcorundum grains. Millan-Gabet et al. (2005) pre-sented the first spatially resolved observations of asample of 23 Mira stars simultaneously measured inthe near-infrared J, H, and K’ bands, using the IOTAinterferometer. For each star they present visibil-ity amplitude measurements as a function of wave-length, revealing the general relation: J diameter <H diameter < K’ diameter.

Ireland et al. (2005) performed optical inter-ferometric polarimetry measurements of the Mira-like variables R Car and RR Sco, using the SUSI ar-ray. By making visibility measurements in two per-pendicular polarizations, the relatively low-surfacebrightness light scattered by atmospheric dust couldbe spatially separated from the bright Mira photo-spheric flux. This was the first reported successfuluse of long-baseline optical interferometric polarime-try. Observations were able to place constraints onthe distribution of circumstellar material; the innerradius of dust formation for both stars was foundto be less than 3 stellar radii: much closer thanthe expected innermost stable location for commonly

assumed astrophysical ”dirty silicate” dust (silicatedust with a significant iron content) in these systems.Fedele et al. (2005) presented near-infrared K-bandinterferometric measurements of the Mira star RLeonis obtained with the VLTI/VINCI. These mea-surements indicate a center-to-limb intensity varia-tion that is clearly different from a uniform disc in-tensity profile. Also, they showed that these mea-sured visibility values are consistent with predic-tions from recent self-excited dynamic Mira modelatmospheres that include molecular shells close tocontinuum-forming layers. Perrin et al. (2005) re-ported IOTA/FLUOR interferometry with narrowspectral bands, isolating the near-continuum andstrong molecular features, obtained for the super-giant µ Cep. Their model shows that a stellar pho-tosphere of angular diameter 14.11 ± 0.60 mas issurrounded by a molecular layer of diameter 18.56 ±0.26 mas

Boboltz et al. (2005) presented the first co-ordinated Very Long Baseline Array (VLBA)/VeryLarge Telescope Interferometer (VLTI) measure-ments of the stellar diameter and circumstellar atmo-sphere of a Mira variable star. Observations of theSiO maser emission toward the Mira variable S Oriwere conducted using the VLBA, while near-infraredK-band measurements of the stellar diameter wereperformed using VLTI/VINCI and closely spaced intime to the VLBA observations. Their measure-ments show that the masers lie relatively close to thestellar photosphere at a distance of ∼2 photosphericradii, consistent with model estimates. Weiner etal. (2006) performed mid-infrared observations ofIK Tau at 11.15 µm with the three-telescope ISI,and also using individual segments of the Keck tele-scope for multiple-aperture interferometry at 10.7µm. Both experiments provided closure phase andshow temporal variations and asymmetries in thesurrounding dust, with a difference of about 15 % inintensity between two sides of the star. Comparisonwith earlier interferometric measurements showed asubstantial reduction in dust surrounding the starover one decade. Interferometric observations of sixMira-type stars: R Aqr, CIT 3, χ Cyg, W Aql, RLeo, and U Ori have been performed by Tatebe et al.(2006) at Infrared Spatial Interferometer (ISI) whichwas comprised of three telescopes. The deduced one-dimensional images show significant changes in thestars and surrounding dust between the two consec-utive years, indicating a non-constant gas emission.

Wittkowski et al. (2007) presented the firstmulti-epoch study that includes mid-infrared and ra-dio interferometry of the oxygen-rich Mira star SOri obtained with VLTI/MIDI at four epochs andmaser emission obtained with the VLBA at threeepochs. They concluded that S Ori shows significantphase-dependences of photospheric radii and dustshell parameters and that the Al2O3 dust grains andSiO maser spots form at relatively small ∼ 1.8-2.4photospheric radii. Their results suggest increasedmass loss and dust formation close to the surfacenear the minimum visual phase when Al2O3 dustgrains are co-located with the molecular gas andthe SiO maser shells, and a more expanded dustshell after visual maximum. Ohnaka et al. (2007)

13

S. JANKOV

carried out VLTI/MIDI observations of carbon-richMira variable V Oph at three different phases withthree different baselines using four 8.2 m Unit Tele-scopes. Their observations and modeling indicatethat carbon-rich Miras also have extended layersof polyatomic molecules as previously confirmed inoxygen-rich Miras. The temporal variation of theN-band angular size has been found to be largelygoverned by the variations of the opacity and the ge-ometrical extension of the C2H2 layers and the dustshell, which masks the size variation of the photo-sphere.

Using a grism at the Keck Interferometer, Eis-ner et al. (2007a) obtained spectrally dispersed (R ∼230) interferometric measurements of the Mira starR Vir. Their data showed that the measured radiusof the emission varies substantially from 2.0 to 2.4µm and simple models could reproduce these varia-tions using extended molecular layers which absorbstellar radiation and reemit it at longer wavelengths.As they observed spectral regions with and withoutsubstantial molecular opacity, they could determinethe stellar photospheric radius, uncontaminated bymolecular emission. They infer that most of themolecular opacity arises at approximately twice theradius of the stellar photosphere.

Asymmetries and motions in the dust shellsurrounding o Ceti have been reported by Chan-dler et al. (2007). The measurements were takenwith the ISI, a three-element interferometer operat-ing at 11.15 µm and three years of data permiteddetection of the movement of dust shells in time.Wittkowski et al. (2008) presented J, H, K spec-trally dispersed interferometry of S Ori with a spec-tral resolution of 35 for the Mira variable S Ori-onis, with the VLTI/AMBER instrument between1.29 µm and 2.32 µm. The measured visibility anduniform disc diameter variations with wavelength re-semble and generally confirm the predictions by pre-vious dynamic model atmospheres. These size vari-ations with wavelength could be understood as theeffects arising from water vapor and CO layers lyingabove the continuum-forming photosphere.

The sizes and shapes of the stars o Ceti andR Leonis have been measured in the mid-infrared byTatebe et al. (2008) using the ISI instrument. Thestar o Ceti appeared to be rather symmetric whilethe shape of the R Leonis appeared more consistentwith a uniform disc plus a point source. Pluzhniket al. (2009) presented the results of long baselineinterferometric observations of the Mira-type starU Ori at 1.51 µm (H2O band), 1.64 µm (pseudo-continuum), and 1.78 µm (H2O band), using thethree-element IOTA interferometer. They performedmodel-independent image reconstruction of the enve-lope around the star using measured visibilities andclosure phases. The images show asymmetric struc-ture of the U Ori envelope (Fig. 11), and they dis-cussed the geometric and kinematic structure of theenvelope based on a model of an almost face-on ex-panding and rotating disc around the star.

Le Bouquin et al. (2009a) obtained single-epoch interferometric observations of T Lep with a

continuous dataset in the spectral domain (λ =1.5-2.4 µm) and in the spatial domain (interferometricbaselines ranging from 11 to 96 m), and they per-formed a model independent image reconstructionfor each spectral bin. Reconstructed images (Fig.12) confirm the general picture of a central star par-tially obscured by the surrounding molecular shell ofchanging opacity. At 1.7 µm, the shell becomes opti-cally thin, with corresponding emission appearing asa ring circling the star. This was the first direct ev-idence of the spherical morphology of the molecularshell in Mira stars while model fitting confirmed aspherical layer of constant size and changing opacityover the wavelengths.

Fig. 11. Reconstructed images of U Ori. Imagesare reconstructed at 1.51 µm, 1.64 µm, and 1.78 µm(top, from left to right). The corresponding imagesof the envelope, obtained by removing the model ofthe central source are shown in the second row. Thebright spots A, B, and C correspond to H2O maserfeatures. The synthesized beams of the interferome-ter for all wavelengths are shown at the bottom.

Fig. 12. Reconstructed images of T Lep. Imagesare shown for several AMBER spectral bins acrossthe H and K bands. The interferometric beam sizeis displayed at the bottom-right part of each image.Spatial scale is in mas.

14

OPTICAL INTERFEROMETRY II

Lacour et al. (2009) presented infrared inter-ferometric imaging of the S-type Mira star χ Cygni.The star was observed at four different epochs withthe IOTA interferometer (H band) using the inte-grated optics recombiner IONIC. Images show up to40 % variation in the stellar diameter, as well assignificant changes in the limb darkening and stel-lar inhomogeneities. The model fitting gave precisetime-dependent values of the stellar diameter, andreveals presence and displacement of a warm molec-ular layer. The constant acceleration of the COmolecules during 80 % of the pulsation cycle leadto argument for a free-falling layer.

In order to explore the photosphere of thevery cool late-type star VX Sgr and, in particu-lar, the characterization of molecular layers abovethe continuum forming photosphere, Chiavassa etal. (2010) obtained interferometric observations withthe VLTI/AMBER instrument. Reconstructed im-ages and visibilities showed a strong wavelengthdependence and the H-band images displayed twobright spots whose positions were confirmed by thegeometrical model.

9. PLANETARY NEBULAE

After having been extensively studied in thevisible, the Planetary Nebulae and among them theyoungest and more dusty ones are now studied in thenear and mid-infrared. It is commonly accepted thatdiscs surrounding the central star can be an essentialingredient to the shaping of planetary nebulae, butthe spatial resolution of single aperture astronomicalinstruments is usually not sufficient for detecting andstudying these discs. The geometry of the disc andthe mass stored are the key parameters for constrain-ing the models of nebula formation, and for tracingback the evolution of the central star. The disc in-ner edge, as seen from the star, could be thick anddense enough to collimate stellar winds into lobesand knowing its geometry it enables us to better un-derstand the distribution of illuminated and shad-owed regions in the extended nebula. The significantprogress for these studies has been achieved with theVLT Interferometer and its two instruments: AM-BER operating in the near-infrared and MIDI in themid-infrared providing a typical spatial resolution of2 and 10 mas, respectively, which is well suited forthe study of many aspects of the late evolution ofstars and in particular, to deal with the asymmetryin the Planetary Nebula, back to the AGB stars.

Chesneau et al. (2006) presented high spatialresolution observations of the dusty core of CPD-56 8032, taken with the mid-infrared interferometerVLTI/MIDI. The infrared core was almost fully re-solved with the three baselines and the signal is in-terpreted in terms of a ring structure which woulddefine the bright inner rim of the equatorial disc.Geometric models allowed derivation of the main ge-ometrical parameters of the disc, showing that thedisc is mostly optically thin in the N band and highlyflared. Lagadec et al. (2006) reported on infraredobservations of the planetary nebula Hen 2-113 ob-

tained with the same instrument. No clear core at8.7 µm and no fringes through the N band couldbe detected and a qualitative interpretation of theobject structure is proposed by using a diabolo-likegeometrical model. In order to obtain a high spa-tial resolution information on the dusty core of bipo-lar planetary nebulae and to directly constrain theshaping process, Chesneau et al. (2007a) obtainedobservations of the dusty core of the extreme bipo-lar planetary Ant nebula (Mz 3, Hen 2-154) takenwith the VLTI/MIDI. The core was clearly resolvedand they suggest an edge-on disc whose major axisis perpendicular to the axis of the bipolar lobes forwhich they deduce inclination and position angle.

With the purpose of studying the structure ofthe nebula around the post AGB, the binary star 89Her, Bujarrabal et al. (2007) obtained N-band inter-ferometric data on the extent of the hot dust emissionobtained with the VLTI/MIDI instrument and theIRAM Plateau de Bure Interferometer (PdBI) ob-servations of the 12CO(1-0) and 12CO(2-1) lines andcontinuum emission. They presented high-resolutionPdBI maps, and derived the properties of the neb-ula from model fitting. Two nebular components arefound: an extended hour-glass-like structure and anunresolved very compact component, smaller than0.4 arcsec. Assuming that the compact componentis a Keplerian disc, they derive disc properties thatare compatible with expectations for such a struc-ture; in particular, the size of the rotating gas discthat is very similar to the extent of the hot dust com-ponent from VLTI data. Assuming that the equatorof the extended nebula coincides with the binary or-bital plane, they provide new results on the compan-ion star mass and orbit. In order to probe the in-nermost region of the circumstellar dust shell of thedeeply embedded Wolf Rayet star WR 118, Millouret al. (2009a) carried out spectro-interferometric ob-servations using the VLTI/AMBER instrument inthe low-spectral resolution mode (R=35). The K-band observations were obtained with three 1.8 mtelescopes spanning projected baselines between upto 40 m which allowed them to conclude that WR118 possibly harbors a pinwheel nebula, suggestinga binary nature of the system.

10. LUMINOUS BLUEVARIABLES: η Carinae

The bright star η Carinae is the most mas-sive and luminous star in our region of the MilkyWay. Though it has been extensively studied usingmany different techniques, its physical nature andthe mechanism that led to the creation of the Ho-munculus nebula are still debated. Van Boekel etal. (2003) presented new high angular resolution ob-servations at near-IR wavelengths of the core of thenebula by using the VLTI/VINCI instrument. Theobservations provided spatial information on a scaleof 11 AU at the distance of η Carinae and the datashow that the object is elongated with a de-projectedaxis ratio of approximately 1.5 and that the majoraxis is aligned with that of the nebula. The most

15

S. JANKOV

likely explanation for this observation was a model inwhich stellar rotation near the critical velocity causesenhanced mass loss along the rotation axis resultingfrom the large temperature difference between thepole and equator in rapidly rotating stars. They con-clude that η Carinae must rotate in excess of 90% ofits critical velocity to account for the observed shapeand that the large outburst may have been shapedin a similar way.

To constrain spatially and spectrally the warmdusty environment and the central object, Ches-neau et al. (2005b) observed the core of the neb-ula surrounding η Carinae with the interferom-eter VLTI/MIDI. In particular, narrow-band im-ages at 3.74 µm and 4.05 µm reveal the butterflyshaped dusty environment close to the central starwith unprecedented (sub-arcsecond) spatial resolu-tion (Fig. 13). A void whose radius corresponds tothe expected sublimation radius has been discoveredaround the central source.

Fig. 13. The butterfly shaped dusty environmentclose to η Car. Zoom into the deconvolved imagesfrom the 3.74 µm (top) and 4.05 µm (bottom) fil-ters. The raw images are shown on the left side, thedeconvolved images at iteration 10 and 40 are shownon the middle and on the right side.

To study the wavelength dependence of theη Carinae optically thick wind region with a highspatial resolution of 5 mas (11 AU), Weigelt et al.(2007) presented the first NIR spectro-interferometryof the star. The analysis has been performed on theVLTI/AMBER spectrally dispersed interferograms(spectral resolutions of 1500 in the medium spec-tral resolution mode and 12000 in the high resolu-tion mode) which allowed the investigation of thewavelength dependence of the visibility, differentialphase, and closure phase. For the interpretation ofthe non-zero differential and closure phases measuredwithin the Brγ line, they present a simple geometricmodel of an inclined, latitude-dependent wind zone.Their observations supported theoretical models ofanisotropic winds from fast-rotating, luminous hotstars with enhanced high-velocity mass loss near thepolar regions.

With the purpose of resolving the central en-gine of the η Carinae complex in the near-infraredon angular scales of a few milli-arcseconds, Kervella

(2007) used the VLTI/VINCI data in the K bandobtained with either two 0.35 m siderostats or two8-m Unit Telescopes. He reported visibility measure-ments in satisfactory agreement with the previousresults obtained with VLTI/AMBER.

Aiming to constrain the rotational velocity ofthe primary star and to probe the influence of thecompanion, Groh et al. (2010) analyzed K-bandcontinuum visibilities from VLTI/VINCI and clo-sure phase measurements from VLTI/AMBER. Theyconclude that the density structure of the primarywind can be sufficiently disturbed by the compan-ion, thus mimicking the effects of fast rotation, andtherefore the fast rotation may not be the only ex-planation for the interferometric observations.

11. NOVAE

The high spatial resolution studies of Novaein the optical domain has been recently stimulatedby emergence of large optical interferometers. Par-ticularly, the outburst that occurred on February 12,2006 on recurrent nova RS Oph has been extensivelyobserved by different interferometers. Following theoutburst, Monnier et al. (2006a) measured its near-infrared size using the IOTA, Keck, and PTI inter-ferometers at multiple epochs. They concluded thatthe characteristic size of∼ 3 mas hardly changed overthe first 60 days of the outburst, ruling out popularmodels where the near-infrared emission arises fromhot gas in the expanding shock. The emission wasalso found to be significantly asymmetric, evidencedby nonzero closure phases detected by IOTA. Usingthe PTI interferometer, Lane et al. (2007) performedobservations of RS Oph, resolving the emission fromthe nova for several weeks after the outburst. Theyreported that the near-IR source which initially ex-panded to a size of ∼ 5 mas, began to shrink aroundday 10, and reached ∼ 2 mas by day 100. They ar-gue that the fact that the emission region appears toshrink does not necessarily imply that the material isfalling; what one can see is the effective photosphericdiameter, which is a function of the density of thematerial, and if the mass-loss rate from the centralwhite dwarf drops, the apparent photospheric radiuswould be expected to shrink even as the materialcontinues flowing outward.

Chesneau et al. (2007b) reported on spec-trally dispersed VLTI/AMBER observations of RSOph five days after the discovery of the outburst.Using three baselines and a spectral resolution ofR=1500, they measured the extension of the milli-arcsecond scale emission in the K band continuumand in the Brγ and He I 2.06 µm lines, which al-lowed to get an insight into the kinematics of theline forming regions. Their results confirm the basicfireball model (Fig. 14), contrary to the conclusionsof other interferometric observations conducted byMonnier et al. (2006a). Barry et al. (2008) reportedN-band (8 to 12.5 µ) observations of RS Oph usingthe Keck Interferometer 3.8 days following the out-burst. The data show evidence of enhanced neutralatomic hydrogen emission and atomic metals includ-

16

OPTICAL INTERFEROMETRY II

ing silicon located in the inner spatial regime near thewhite dwarf. They report also nebular emission linesand evidence of hot silicate dust in the outer spatialregion, centered at ∼ 17 AU from the white dwarf,which were not found in the inner regime. Theirresults support a model in which the dust appearsto be present between outbursts and is not createdduring the outburst event.

Fig. 14. RS Oph outburst. Sketch of the near-IRellipses extensions compared with the radio structureobserved at t= 13.8d (thick extended ring, O’Brienet al. 2006). The continuum ellipse is delimited bythe solid line, the ellipse that corresponds to the coreof Brγ by the dotted line and the one correspondingto the core of HeI by the dashed line. The small dot-ted line delimits the Brγ ellipse scaled at t= 13.8 d.North is up, east left.

Lane et al. (2007) have resolved the classicalnova V1663 Aql using a long-baseline near-IR inter-ferometry from the PTI interferometer covering theperiod from ∼ 5 to 18 days after the peak bright-ness. They directly measured the shape and size ofthe fireball, (which they found to be asymmetric)and the apparent expansion rate. Assuming a linearexpansion model, the time of the initial outburst ap-proximately 4 days prior to the peak brightness wasinfered. In order to improve the distance determina-tion and to constrain the mechanisms leading to veryefficient dust formation under the physical conditionsencountered in novae ejecta, Chesneau et al. (2008)presented the first high spatial-resolution monitoringof the dust-forming nova V1280 Sco. Spectra and vis-ibilities were regularly acquired between the onset ofdust formation, 23 days after the discovery (11 daysafter the maximum), using the beam-recombiner in-struments AMBER (near-IR) and MIDI (mid-IR).These observations allowed determination of the ap-parent linear expansion rate for the dust shell 0.35± 0.03 mas day, and the approximate ejection time

of the matter in which dust formed (tejec = 10.5± 7d), i.e. close to the maximum brightness.

Chesneau et al. (2011) reported on near-IRinterferometric observations of the recent 2011 out-burst of the recurrent nova T Pyx. They obtainednear-IR observations of T Pyx at dates ranging from2.4 to 48.2 days after the outburst with the CLASSICrecombiner located at the CHARA array and withthe PIONIER and AMBER recombiners located atthe VLTI array. Comparing expansion of the H andK band continua as well as the Brγ emission line,and infering information on the kinematics and mor-phology of the early ejecta, they concluded that itcould be most simply interpreted within the frameof a bipolar model oriented nearly face-on.

12. YOUNG STELLAR OBJECTS

The very close environments of young stars arethe hosts of fundamental physical processes such asstar-disc interactions, mass accretion, and ejection.The extraordinary interest in the inner region of acircumstellar disc results from the assumption thatthe formation of planets is favored there, focusing onhow accretion discs evolve into protoplanetary discsand, finally, to debris discs and planets. Optical in-terferometry is playing an important role in eluci-dating the earliest stages of planetary formation byprobing the density and temperature structure of thediscs presumably before the planet formation. Thefirst successful observation of a Young Stellar Ob-jects (YSOs) with stellar interferometry was reportedby Malbet et al. (1998) who resolved the young out-bursting star FU Ori using a 100-m long baseline inthe K band with the PTI. Using the IOTA interfer-ometer Millan-Gabet et al. (1999) resolved the Her-big Ae/Be star AB Aur, finding the near-IR emissionto be much larger than expected.

Radically improved infrared detectors have re-cently allowed optical interferometers to investigatethe inner accretion discs around young stellar objects(Millan-Gabet et al. 2001, Akeson et al. 2000, 2002,Tuthill et al. 2001, Danchi et al. 2001). In mostcases, the near-IR sizes were found to be significantlylarger than expected from the favored disc models ofthe time (Malbet and Bertout 1991). The complexmorphological structure of these environments hasbeen confirmed by the now quite rich data sets ob-tained by near-infrared long-baseline interferometry.While observations in millimeter range probe coolerouter disc regions and layers close to the midplane ofcircumstellar discs, observations in the mid-infraredwavelength regime are more sensitive to warmer discarreas, such as the surface of the inner regions wherethe dust is directly irradiated by the central star.

12.1. Herbig Ae/Be stars

Pre-main sequence stars in the intermediatemass range (1.5-3 M¯), called Herbig Ae and Bestars, are observed to be surrounded by circumstel-lar material which reveals itself by discrete emission

17

S. JANKOV

lines and by continuous excess emission in spectralenergy distribution (SED). The spatial distributionof this material however has been subject to de-bate: both geometrically flat disc models and spher-ically symmetric envelope mo