Serb. Astron. J. } 181 (2010), 1 - 17 UDC 520.3613DOI: 10.2298/SAJ1081001J Invited review

ASTRONOMICAL OPTICAL INTERFEROMETRY.I. METHODS AND INSTRUMENTATION

S. Jankov

Astronomical Observatory Belgrade, Volgina 7, 11060 Belgrade 38, SerbiaEmail: sjankov@aob.rs

(Received: November 17, 2010; Accepted: November 17, 2010)

SUMMARY: Previous decade has seen an achievement of large interferometricprojects including 8-10m telescopes and 100m class baselines. Modern computerand control technology has enabled the interferometric combination of light fromseparate telescopes also in the visible and infrared regimes. Imaging with milli-arcsecond (mas) resolution and astrometry with micro-arcsecond (as) precisionhave thus become reality. Here, I review the methods and instrumentation cor-responding to the current state in the field of astronomical optical interferometry.First, this review summarizes the development from the pioneering works of Fizeauand Michelson. Next, the fundamental observables are described, followed by thediscussion of the basic design principles of modern interferometers. The basic inter-ferometric techniques such as speckle and aperture masking interferometry, aperturesynthesis and nulling interferometry are disscused as well. Using the experience ofpast and existing facilities to illustrate important points, I consider particularly thenew generation of large interferometers that has been recently commissioned (mostnotably, the CHARA, Keck, VLT and LBT Interferometers). Finally, I discussthe longer-term future of optical interferometry, including the possibilities of newlarge-scale ground-based projects and prospects for space interferometry.

Key words. Instrumentation: interferometers Methods: observational Tech-niques: high angular resolution

1. INTRODUCTION

An optical (visible and infrared) long-baselineinterferometer is a device that allows astronomers toachieve a higher angular resolution than it is possi-ble with conventional telescopes. In fact, at wave-length , the resolution of a single telescope withaperture diameter D scales as D/ while the reso-lution of two-telescope interferometer scales as B/,where the baseline B is the distance between the tele-scopes. However, more severely, both are limited bythe atmospheric turbulence (seeing typically 1 arc-second). For this reason, a single aperture telescope

can be used as an interferometric device, but usuallyit is composed of an array of at least two telescopes,which sample the wavefronts of light emitted by asource at separate locations, and redirect starlight toa central location in order to recombine the sampledwavefronts and to produce interference fringes. Thecontrast of interference fringes, or visibility, variesaccording to the characteristics of the light source(for example, the size of a star or the separation be-tween two stars in a binary system) and accordingto the length and orientation of the interferometersbaseline, the line connecting the two telescope aper-tures. It is possible to take measurements from manydifferent baselines, most easily by waiting while the

1

S. JANKOV

Earth rotates. In addition, most of the new inter-ferometers have more than two mirrors in the arrayand can move the apertures along tracks.

The implementation of interferometry in opti-cal astronomy began more than a century ago withthe work of Fizeau, who outlined basic concept ofstellar interferometry, how interference of light canbe used to measure the sizes of stars. He suggestedthat one could observe stars using a mask with twoholes in front of a large telescope and that it shouldbe possible to obtain some information on the angu-lar diameters of these stars (Fizeau 1868). The firstattempts to apply this technique were carried outsoon thereafter (Stephan 1874), although the largestreflecting telescope in existence at the time, the 80cmreflector at the Observatoire de Marseille, could notresolve any stars with a mask that had two holes sep-arated by 65cm, and only the upper limit (0.158 arc-sec) of stellar diameter could be derived. However,with a Fizeau interferometer on the 1m Yerkes refrac-tor, Michelson (1891a,b) measured the diameters ofJupiters Galilean satellites. Following Michelsonssuccess Schwarzschild (1896) managed to resolve anumber of double stars with grating interferometeron 25cm Munich Observatory telescope, while notlong afterward, on the 2.5m telescope on Mt. Wilson,Anderson (1920) determined the angular separationof spectroscopic binary star Capella ( Aur).

Although the first measurement of a stellar an-gular diameter was performed on the supergiant starBetelgeuse ( Orionis) with Michelsons 6m stellarinterferometer in 1920 (Michelson and Pease 1921),the optical interferometry was slowly evolving from adifficult laboratory experiment to a mainstream ob-servational technique. Following the success of the6m interferometer, Pease (with Hale) constructed a15m interferometer but this experiment was not verysuccessful. Due to the disappointing results from the15m interferometer, it would be decades before sig-nificant developments inspired new activity in theoptical field. The real difficulty is to combine thebeams in phase with each other after they have tra-versed exactly the same optical path from the sourcethrough the atmosphere, each telescope, and furtherto the beam recombination point. This has to bedone to an accuracy of a few tenths of the wave-length, which in the case of visible light, is not an ob-vious task, particularly because of atmospheric tur-bulence which makes the apparent position of a staron the sky jitter irregularly. This jitter often causesthe beams in different arms of the interferometer tooverlap imperfectly or not at all at any given mo-ment. For these reasons, the optical interferome-try requires extreme mechanical stability, sensitivedetectors with good time resolution, and at least asimple adaptive optical system to reduce the effectsof atmospheric turbulence. For these reasons, onlythe technologies emerging at the end of 20th centuryallowed the full application of optical interferometryin astronomy.

Meanwhile, advances in radar technique dur-ing World War II stimulated rapid development ofradio interferometry beginning with the first radiointerferometer built by Ryle and Vonberg (1946).The main reason for tremendous success concern-

ing astronomical applications is that the effects ofatmospheric noise are much less pronounced at ra-dio wavelengths. Paralelly, the development of In-tensity interferometry discovered by Hanbury Brownand Twiss (1956a) inspired a new project in opticalinterferometry. Their basic principle describes howcorrelations of intensities (not electric fields) can beused to measure stellar diameters. The importantpoint is that the technique relies on the correlationbetween the (relatively) low-frequency intensity fluc-tuations at different detectors, and that it does notrely on the relative phase of optical waves at thedifferent detectors. The requirements for the me-chanical and optical tolerances of an intensity in-terferometer are therefore much less stringent thanin the case of direct detection schemes as it isFizeau/Michelson interferometery. First results werereported soon thereafter (Hanbury Brown and Twiss1956b), leading to the development of the Narrabriintensity interferometer. With a 188m longest base-line and blue-sensitivity, this project had a profoundimpact on the field of optical interferometry, mea-suring dozens of hot-star diameters (e.g. HanburyBrown et al. 1967a,b, 1970, 1974a,b, Davis et al.1970). The small bandwidths attainable with in-tensity interferometry limited the technique to thebrightest stars, and pushed the development of so-called direct detection schemes, where the light iscombined before detection to allow large observingbandwidths.

With the advent of lasers in visible and in-frared, the benefits of radio interferometry havebeen pursued in optical long-baseline interferometrythrough heterodyne detection (e.g. Gay and Journet1973, Assus et al. 1979). Radio interferometry func-tions in a fundamentally different way from opticalinterferometry. Radio telescope arrays are hetero-dyne, meaning that incoming radiation is interferedwith a local oscillator signal before detection. Thesignal can then be amplified and correlated with sig-nals from other telescopes to extract visibility mea-surements. Optical interferometers are traditionallyhomodyne, meaning that incoming radiation is in-terfered only with light from other telescope. Thisrequires transport of the light to a central station,without the benefit of being able to amplify the sig-nal. The other important benefit of radio interfer-ometry is that the required accuracy for beam re-combination is much more easy to achieve in radiothen in optical domain. One heterodyne optical in-terferometer (ISI, see Section 3.1) has been built tooperate at 10m wavelengths. As for Intensity in-terferometry, the technique is feasible but with smallbandwidths attainable and limited to bright sources,while the homodyne optical interferometry allowslarge bandwidths to be used since the interfered lightis detected directly. Two important steps towardsmodern optical interferometry have been done at theObservatoire de Calern, France:

1. Labeyrie (1970) proposed speckle inter-ferometry, a process that deciphers the diffraction-limited Fourier spectrum and image features of stel-lar objects by taking a large number of very-short-exposure images of the same field. This techniquecan be applied to reconstruct a single image that is

2

OPTICAL INTERFEROMETRY I

free of turbulent areas, in essence an image of theobject as it might appear free of atmospheric turbu-lence.

2. Long-baseline interferometry on a star bydirectly combining the electric fields before photondetection (direct detection) was first accomplishedin 1974 by Labeyrie (1975) by using a 12m baselineat the I2T interferometer.

The following section deals with the funda-mental observables in astronomical optical interfer-ometry. In Section 3. three basic designs to achievecoherent combination of light are explained in moredetails, while in Section 4. the basic interferometrictechniques are presented. Further development inthe field is described in Sections 5. and 6. while thepossibilities of new ground-based and space projectsare discussed in Section 7. Finally, I summarize themost important conclusions in Section 8.

2. INTERFEROMETRIC OBSERVABLES

2.1. Visibility modulus

The fringe contrast observed with an interfer-ometer is historically called the visibility and, for thesimple (two-slit) interferometer can be written as:

V =Imax IminImax + Imin

=Fringe amplitudeAverage intensity

(1)

where Imax and Imin denote the maximum and min-imum intensity of the fringes.

Generally, the primary observable of an inter-ferometer is the complex visibility function (the am-plitude and phase), which is the Fourier transformof the object brigthness distribution (van Cittert-Zernike Theorem). Let I(u, v, ) denote the spa-tial Fourier transform of sky brightness I(s, p, ) atthe wavelength , where the Fourier frequency pairu, v (spatial frequency) is the conjugate of the spa-tial Cartesian coordinates s, p, respectively. Defin-ing the complex visibility V (u, v, ) = I(u, v, )/Nwhere N is normalization factor defined so that|V (0, 0, )| = 1, it follows:

V = |V (u, v, )|.

The visibility V as defined by Eq. (1) is exactlyproportional to the amplitude modulus of the imageFourier component corresponding to the spatial fre-quency (u, v) = ~b/ where ~b is the baseline vector ~Bprojected onto the plane of the sky.

The phase of the fringe pattern is equal tothe Fourier phase of the same spatial frequency com-ponent but it cannot be observed since an incomingplane wave from a source is corrupted as it propa-gates through the turbulent atmosphere which pro-duces atmospheric phase delays such that the phaseinformation on the source is completely lost. Prac-tically only the modulus of the complex visibilityV = |V (u, v, )| (or squared visibility amplitude V 2)is an observable quantity.

The corruption of this phase informationhas serious consequences, since imaging of non-centrosymmetric objects rely on the Fourier phaseinformation encoded in this intrinsic phase of inter-ferometer fringes. Without this information, imag-ing cannot be done except for simple objects such asdiscs or round stars. A number of strategies havebeen developed to circumvent these difficulties.

2.2. Relative phase

The technique called phase referencing relieson the fact that the relative phase can be obtainedif atmospheric and instrumental noise can be con-trolled or compensated, for example through ref-erencing by measuring the instantaneous phases offringes using a reference star (Shao and Colavita1992) or by using the same source at another wave-length. The technique, called Differential SpeckleInterferometry (DSI) measures the relative shift ofan object in different spectral bands, and its appli-cation is based on the assumption that the shift be-tween two speckle images (instantaneous distortedimages, see Section 4.1) can be related to spatialstructures of the object under study. Althoughthe phase of the spatial coherence function is cor-rupted by instrumental and atmospheric noise, dif-ferent methods of data processing can be used toovercome this difficulty. If the two sets of speckleinterferograms of the same object are recorded si-multaneously at two different wavelengths, but closeenough for the Point Spread Function to be the samefor both channels, then their ensemble average cross-spectrum provides the information about the rela-tive shift as a fraction of the object size. This tech-nique was described by Beckers (1982), Jankov et al.(2001).

2.3. Closure phases

The closure phase measurements tell us aboutthe symmetry of an object and is only available froma closed triangle of elements, such as three tele-scopes. Normally, atmospheric fluctuations disturbthe fringe phase between any two interferometer el-ements, but these fluctuations cancel out in a closedtriangle.

Considering a 3-telescope array where thephases ij, jk, ki are intrinsic phases, ij,

jk,

ki

are phases measured on the baselines between tele-scopes (i and j), (j and k), (k and i) respectively, i,j, k are the atmospheric errors introduced abovetelescopes (i,j,k) and ij, jk, ki is the noise, it fol-lows:

ij = ij + i j + ijjk = jk + j k + jkki = ki + k i + ki

Consequently the closure phase:

cl = ij+jk+

ki = ij+jk+ki+ij+jk+ki

is free of atmospheric phase noise (Jennison 1958).

3

S. JANKOV

2.4. Bispectrum

When analysing the speckle interferometrydata, one can think of many virtual sub-apertures(with sizes equal to the coherence length 2 ) spreadacross the full telescope, with fringes forming be-tween all the sub-aperture pairs. The speckle in-terferometry data can be reduced using the bispec-trum, which permits a direct inversion from the es-timated Fourier amplitudes and phases. The bispec-trum:

Bijk = VijVjkVkiis formed through triple products of the complex vis-ibilities around a closed triangle, where i, j, k specifiesthe three aperture locations on the pupil of the tele-scope.

It was discovered that the Fourier phasescould also be estimated from such data (e.g. Knoxand Thompson 1974), and that the bispectrumphase is identical to the closure phase. Interest-ingly, the use of the bispectrum for reconstruct-ing diffraction-limited images was developed inde-pendently (Weigelt 1977) of the closure phase tech-niques, and the connection between the approacheswas realized only later (Roddier 1986, Cornwell1987).

3. INSTRUMENTAL DESIGNS

3.1. Heterodyne interferometer

The principle of a Heterodyne interferometeris to change the frequency of light at each telescope,carry to the common focus an intermediate frequencysignal, and combine all these signals in order to pro-duce fringes. Associated to each telescope is a laser,and the two beams (laser and star) are then com-bined with a beam splitter onto an infrared-sensitivephotodiode, which produces an electric current pro-portional to the light intensity on it. Two lightwaves, those from the laser and from the star, arevery close in frequency and while combining twobeams, they beat together to produce a signal atlower frequency such that it becomes a radio wave.

To produce interference fringes two radio sig-nals are overlapped instead of light beams. All ofthe information that is carried in the visible or in-frared light wave from the star is still in the newradio signal, the only thing that is different is thewavelength. The principle of heterodyne interferom-etry is depicted in Fig. 1.

This method of converting the frequency ofa signal from the star by mixing it with a fixed-frequency local oscillator is called heterodyne de-tection. (The term local oscillator refers to thelaser in this case). Usually, the local oscillatorsare CO2 lasers operating at frequencies 10mwavelengths. The heterodyne detection suffers frombandwidth limitations as well as additional noisecontribution from laser shot-noise, which becomesprogressively worse at higher frequencies. However,one of the main advantages is that the signals areconverted to radio frequencies and then combinedelectronically using techniques from radio astronomy.

Fig. 1. Heterodyne interferometer arrangement.The light from an infrared laser together with infraredlight from a star are converted to radio frequenciesthrough heterodyne circuits and then combined elec-tronically. A light wave from a star hits first tele-scope, and has to travel an extra distance (d) beforeit arrives at second telescope. In order for the in-terference method to work properly, the signal fromone telescope should be delayed in time () until thewavefront reaches the other telescope; only then twosignals can interfere together.

3.2. Intensity interferometer

The Intensity interferometer is also called thepost-detection correlation Interferometer since thesignals are correlated after detection. The inten-sity interferometry technique does not rely on theactual interference of light rays, thus no fringes areformed. Instead, the interferometric signal, the de-gree of mutual coherence, is characterized by the de-gree of correlation of light intensity fluctuations ob-served at two different detectors. In practice, the de-gree of mutual coherence is measured using fast tem-poral correlations between narrow optical wavebandintensity fluctuations observed by two (or more) tele-scopes separated by a baseline distance (Fig. 2).

Fig. 2. Intensity interferometer arrangement. Theinterferometric signal, the degree of mutual coher-ence, is characterized by the degree of correlation oflight intensity fluctuations observed at two differentdetectors.

4

OPTICAL INTERFEROMETRY I

The principal component of the intensity fluc-tuation is the classical shot noise which will notdemonstrate any correlation between the two sepa-rated telescopes. The intensity interferometric signalis related to a smaller noise component: the wavenoise. The wave noise can be understood as thebeat frequency in optical intensity between the dif-ferent Fourier components of the light reaching thetelescopes. This wave noise will show correlation be-tween the two detectors, provided there is some de-gree of mutual coherence between the light receivedat the two telescopes.

The technique has the benefit of being muchless sensitive to atmospheric phase fluctuations be-cause the signals input to the correlator have followedroughly the same path through the atmosphere. De-lay tracking and local oscillator stability do not haveto be very accurate. Unlike the direct (Amplitudeinterferometry) the Intensity interferometry is wellmatched to blue/visible operation, and scales to longbaselines with ease (the atmospheric stability re-quirements are a factor of a million less stringent).

The major drawback of Intensity Interferom-etry is that copious quantities of light are necessaryto observe the wave noise signal in the presence ofthe much larger shot noise. Even for brightest stars,very large light collectors ( 100m2 area) are nec-essary to have a sufficient statistical strength to ob-serve non-Poisson fluctuations around the mean pho-ton intensity. Mainly for this reason, the experimentconducted at Narrabri from 1963 to 1971 was theonly implementation of a stellar intensity interfer-ometer. Since then, all stellar interferometers havebeen based on Fizeau/Michelsons techniques.

3.3. Direct interferometer

In this approach the light from each apertureis carried to a common focus, combined coherently inorder to detect interferometric signal (fringes). Thecombination of two (or more) light beams can bedone in the image plane (Fizeau mode, Fig. 3a)or in the pupil plane (Michelson mode, Fig. 3b).In the Fizeau mode, the ratio of aperture diame-ter/separation is constant from light collection to re-combination in the image-plane (homothetic pupil).In the Michelson mode, this ratio is not constantsince the collimated beams have the same diameterfrom the output of the telescope to the recombinationlens. This means the object-image relationship canno longer be described as a convolution. However,it is possible to use a relay optics after the beam-combiner so as to stack two output pupils on the topof each other with a modulation depending on theoutput Airy discs distance (Fig. 3c). The main ad-vantage over the Michelson mode recombinations isconservation of the object-image convolution relation(Vakili et al. 2004).

Interferometers in which the output pupil isnot a homothetic image of the input pupil, but thepattern of the subaperture centers is conserved areknown as densified pupil interferometers (Labeyrie1996). In such an arrangement, the interferometer

forms a true image of small sources at the combinedfocus. This is particularly attractive for arrays con-sisting of many telescopes, because the desired infor-mation can be obtained after deconvolution with theknown point spread function.

Fig. 3. Comparison of three different beam-combinations for an optical stellar interferometer.(A) and (B) the classical Fizeau versus Michelsonbeam-combinations, (C) instead of superimposing theAiry patterns from the telescopes, it is possible to usea relay lens after the beam-combiner so as to stackthe two output pupils on the top of each other with amodulation depending on the output Airy discs dis-tance.

The disadvantage of the Michelson mode is avery narrow field of view compared to the Fizeaumode, but for diluted optical arrays it is extremelydifficult to construct an interferometer with imageplane beam recombination. For this reason Fizeaumode is more typically implemented on a common-mount interferometers. Typical example of such aninterferometer designed to create high resolution im-ages over a wide field of view is the Large Binocu-lar Telescope (see Section 6.7), a single-mount struc-ture resulting in a fixed entrance pupil for which theFizeau mode recombination is adapted.

The Michelson mode interferometry is usedin long-baseline interferometers made of independenttelescopes. Beam transfer optics and delay lines al-low it to maintain equal path length between thetelescopes as the object is tracked across the sky.

The real art of developing interferometers isto combine the beams in phase with each other afterthey have traversed exactly the same optical pathfrom the source through each of the telescope downto the beam combination point. The paths are madeequal by adjusting the position of mirrors in the opti-cal delay-line (Fig. 4.) that corrects the drift inducedby the diurnal rotation of the tracked star. Due tothe long path lengths between the telescope and cen-tral beam combining facility, a significant differentialchromatic dispersion occurs if the light is propagat-ing in the air. In order to combine broad bandwidths,one must either transport the light through a vac-uum or construct a dispersion compensator (Tango1990).

5

S. JANKOV

Fig. 4. Michelsons interferometer. The light fromeach aperture is carried to a common focus, combinedcoherently in order to detect interferometric signal(fringes). Beam transfer optics and delay lines al-low it to maintain equal path length between the tele-scopes as the object is tracked across the sky.

In addition, mechanical constraints on the in-strument, errors on the pointing model, thermaldrifts, various vibrations and atmospheric turbulencemake the null Optical Path Difference (OPD) pointchanging. The error on the OPD must be less thanthe coherence length defined by: lc =

2

where is the mean wavelength observed and is thespectral interval. Fringes are searched by adjustingthe delay-line position, and this real-time control iscalled fringetracking.

After being collected by the telescopes andproperly delayed, the light must be directed to acentral facility for beam combination. The interfer-ogram can then be recorded, as long as the scanningtakes place faster than the atmospheric coherencetime c = c/.

4. INTERFEROMETRIC TECHNIQUES

4.1. Speckle Interferometry

Speckle interferometry is a method permit-ting the extraction of spatial information from two-dimensional images at scales down to the diffractionlimit of the telescope, in spite of severe blurring in-troduced by atmospheric turbulence. The image of astar obtained through a large telescope looks speck-led or grainy because different parts of the image areblurred by small areas of turbulence in the earthsatmosphere. The image obtained for an unresolvedpoint source depends on the exposure time. For longexposures, the image is blurred to a seeing disk 1arcsec. For exposures shorter than the coherencetime (tc 10ms for the atmosphere in the optical;100ms for the infrared), a group of bright speckles isobtained approximately of the size of the Airy disk.In that way the speckle interferometry freezes theatmosphere, and gets an instantaneously distortedimage as presented in Fig. 5.

Fig. 5. Speckle, an instantaneously distorted stellarimage.

In 1970, the French astronomer AntoineLabeyrie showed that information could be ob-tained on the high-resolution structure of the objectfrom the speckle patterns using the Fourier analysis(speckle interferometry). In the 1980s, the meth-ods which allowed images to be reconstructed inter-ferometrically from these speckle patterns were de-veloped using a sequence of short-exposure snapshotsto obtain images at a telescopes diffraction limit.

With existing large telescopes, speckle tech-niques thus permit resolution at spatial scales of0.025 arcseconds rather than the 1 to 2 arcsecondsassociated with classical techniques. These meth-ods are also characterized by enhanced measure-ment accuracy of the separation of closely spacedobjects seen through the turbulent atmosphere. Thespeckle interferometry system incorporates an inten-sified charge coupled device array as the primaryimaging detector and an autocorrelator as a highspeed data reduction processor operating at videorates. Also, the reduction of the speckle patterns canbe done a posteriori involving computer processing.

4.2. Aperture Masking Interferometry

Aperture Masking Interferometry is a form ofspeckle interferometry, allowing to reach the maxi-mum possible resolution at ground-based telescopeswith large diameters to produce diffraction limitedimaging.

In the aperture masking technique, the bis-pectral analysis (speckle masking) method is typi-cally applied to data taken through masked aper-tures, where most of the aperture is blocked off andlight can only pass through a series of small holes(subapertures) so that the array of holes acts as aminiature astronomical interferometer. The aperturemask removes atmospheric noise from these measure-ments, allowing the bispectrum to be measured moreprecisely than for an un-masked aperture. For sim-plicity the aperture masks are usually either placedin front of the secondary mirror (e.g. Tuthill et al.2000, see Fig. 6) or placed in a re-imaged apertureplane (e.g. Young et al. 2000).

6

OPTICAL INTERFEROMETRY I

Fig. 6. Desired sparse pupil (left) and the aperturemask used to generate it (right). The required pupilis shown as the set of black spots and is superimposedon a scaled version of the segmented 10m Keck pri-mary mirror (hexagonal segments). From Tuthil etal. (2000).

The masks are usually categorised eitheras non-redundant or partially redundant. Non-redundant masks consist of arrays of small holeswhere two pairs of holes have not the same sepa-ration vector (the same baseline). Each pair of holesprovides a set of fringes at a unique spatial frequencyin the image plane.

Partially redundant masks are usually de-signed to provide a compromise between minimis-ing the redundancy of spacings and maximising boththe throughput and the range of spatial frequen-cies investigated. Although the signal-to-noise ofspeckle masking observations at high light level canbe improved with aperture masks, the faintest limit-ing magnitude cannot be significantly improved forphoton-noise limited detectors so that the principallimitation of the technique is that it is limited torelatively bright astronomical objects.

4.3. Aperture Synthesis

The direct closure phase measurements to-gether with the measurements of visibility amplitudeallow one to reconstruct an image of any object us-ing three or more independent telescopes. This tech-nique has been successfully demonstrated by Bald-win et al. (1996) in the visible band at the Cam-bridge optical aperture-synthesis telescope (COAST,see Section 5.3). Each pair of telescopes in an arrayyields a measure of the amplitude of the spatial co-herence function of the object at a spatial frequency~b/ which corresponds to one point in (u, v) plane.

In order to make an image from an interferom-eter, one needs estimates of the complex visibilitiesover a large portion of the (u, v) plane, both the am-plitudes and phases. The (u, v) points correspondingto a snapshot projection of the baseline are sampledfrom each available baseline. It is also useful to use atechnique called super-synthesis: the (u, v) planeis swept during an observation lasting several hours,due to Earth rotation. After a large variation of hourangle, several visibility moduli are, therefore, measu-

Fig. 7. uv-plane coverage for the observations ofBaldwin et al. (1996). Each point corresponds to anindividual visibility measurement.

red at different positions in (u, v) plane as presentedin Fig. 7.

Reconstruction of complex images involves theknowledge of complex visibilities but the phase of avisibility may be deduced from closure-phase thathas been applied in optical interferometry. The pro-cess after data acquisition consists of phase calibra-tion and visibility phase reconstruction from closure-phase terms by a technique similar to bispectrumprocessing, so that at least three configurations ofthree telescope interferometer is needed. Interferom-eters with two apertures have limited possibilities forimage reconstructions due to absence of the phasevisibility recovering but in that case different spec-tral channels can be employed for differential visibil-ity measurements (between the continuum and spec-tral line) and use of data from one part of the spec-trum such as the continuum to calibrate another part(Jankov et al. 2001, 2002, 2003).

4.4. Nulling Interferometry

This technique, the nulling interferometry,holds a great promise for the detection and char-acterization of Earth-like extrasolar planets at mid-infrared wavelengths, because the light from theparent star arriving on axis is completely rejected(Bracewell 1978). The basic premise of nulling in-terferometry is conceptually quite simple: combinethe starlight that arrives at a pair of telescopes sothat at the centre of the image the two waves areexactly pi out of phase producing a dark centralfringe and effectively making the central star disap-pear. A deep destructive fringe is to be placed acrossthe star, but light coming from sources that are off-set slightly from it (i.e. from planets) gets addedrather that subtracted so that even though the staris completely blanked out, planets at the right loca-tions (near a constructive fringe) are not attenuated(Fig. 8).

An alternative to the nulling Michelson inter-ferometer is the use of a transparent phase-shiftingmask in conjunction with a densified pupil interfer-ometer (Boccaletti et al. 2000). The basic idea of

7

S. JANKOV

Fig. 8. Nulling Interferometry Principle. Thestarlight that arrives at a pair of telescopes so that adeep destructive fringe is placed across the star (mak-ing it disappear), but the light coming from a planetgets added rather that subtracted.

this concept is to introduce a pi phase shift over thecentral part of the point spread function produced bythe densified pupil; a careful adjustment of the sizeof the phase mask together with pupil apodizationin a coronographic setup can in principle provide anexcellent nulling performance.

Nulling interometry can be performed fromthe Earth, especially when imaging Jupiter sizedplanets, but the best place for viewing earth-sizedplanets close to their suns is from a space-based plat-form. Both NASA and ESA have plans to launchinfrared nulling interferometry telescopes into spaceover the next decades (see Section 7).

5. PAST INTERFEROMETERS

In this section, only the most important inter-ferometers that produced significant results in thepast are presented:

5.1. I2T

The I2T (Interferome`tre a` 2 Telescopes), Ob-servatoire Cote DAzur, Plateau de Calern, France(Fig. 9) is the interferometer where the first directinterference fringes between separate telescopes wereobserved and reported by Labeyrie (1975).

Fig. 9. The I2T (Interferome`tre a` 2 Telescopes),Plateau de Calern, France.

I2T made observations in the visible (e.g.,Blazit et al. 1977, Thom et al. 1986) and pioneeredinterferometry at near-IR wavelengths (di Benedettoand Conti 1983, di Benedetto 1985).

5.2. MARK I, II and III

The Massachusetts Institute of Technologyand the Naval Research Laboratory built and op-erated a series of prototype interferometers, namedthe Mark I, Mark II, and Mark III located on Mt.Wilson, California, US.

Active fringe tracking was first demonstratedin 1979 by the Mark I interferometer (Shao andStaelin 1980), while the most popular architecturetoday for the moving delay line is based on the so-lution implemented by the Mark III interferometer(Shao et al. 1988).

The Mark III was specifically designed to per-form wide-angle astrometry, but a variable baselinethat could be configured from 3m to 31.5m providedthe flexibility needed for a variety of astronomicalprograms. Because a full computer control of thesiderostats and delay lines allowed automated acqui-sition of stars and data acquisition, it could observeup to 200 stars in a single night, until 1993, whenthe interferometer stopped the operation.

5.3. COAST

The COAST (Cambridge Optical ApertureSynthesis Telescope) operated from 1991 until 2005at Cambridge University, England. It was plannedas a coherent array of four telescopes operating inthe red and near infra-red, using Michelson interfer-ometry on baselines of up to 100m to give imageswith a resolution down to 1mas.

The number of telescopes (40cm apertures)has been brought up to five, and light from fourof them could be interfered simultaneously. Switch-ing between two four-telescope arrays allowed takingdata on nine baselines and seven closure trianglesduring a single night. Observations could be carriedout in the visible (R,I) or near-infrared (J,H bands).It was the first instrument of its kind to exploit thetechniques of aperture synthesis and closure phaseat optical or infra-red wavelengths, producing thefirst images from an optical aperture synthesis array(Baldwin et al. 1996).

5.4. IOTA

The IOTA (Infrared Optical Telescope Ar-ray) operated from 1993 to 2006 and was an in-terferometer of Smithsonian Astrophysical Observa-tory on Mt. Hopkins, Arizona, US (Fig. 10), withthree 45cm telescopes (enabling closure phase obser-vations) and baselines from 5m up to 38m. Observa-tions have been carried out in the visible or near-IR(J,H,K bands). Visibilities with excellent calibrationcould be obtained with the single-mode fiber systemFLUOR (Fiber Linked Unit for Optical Recombina-tion, Coude Du Foresto et al. 1997) which accountsfor a large fraction of the astronomical results ob-tained with IOTA.

8

OPTICAL INTERFEROMETRY I

Fig. 10. The IOTA (Infrared Optical TelescopeArray) on Mt. Hopkins, US.

5.5. GI2T

The GI2T (Grand Interferome`tre a` 2Telescopes), Observatoire Cote DAzur, Plateau deCalern, France (Mourard et al. 1994, see Fig. 11)was a successor of the I2T since 1985 and used twoboule telescopes with 1.5m apertures on a north-south baseline that could be reconfigured from 10mto 65m.

Fig. 11. GI2T (Grand Interferome`tre a` 2 Telesco-pes), Plateau de Calern, France.

The GI2T had the capability of performingspectrally resolved interferometry. After cloture in2006, its beam combination table, including a ver-satile visible spectrograph and an IR focus has beenmoved to CHARA array (Mourard et al. 2009).

5.6. MIRA

The MIRA (Mitake Infrared Array), NationalAstronomical Observatory, Japan, was an ambitiousplan to build a series of interferometers with in-creasing capabilities. The first phase of this project(MIRA-I), which consisted of two 25cm telescopeswith coude optics on a 4m baseline in Tokyo has suc-cessfully been completed in 1998, and has acquired

stellar fringes. The next step was to built an instru-ment with a slightly larger siderostats and a 30mbaseline (MIRA-I.2). The interferometer ceased op-eration in 2007.

5.7. PTI

The PTI (Palomar Testbed Interferometer),NASA JPL, Mt Palomar, CA, US (Colavita et al.1999, see Fig. 12) was built in 1995 and it consistedof three 40cm siderostats, up to 110m baselines whichprovided 3mas resolution in the near-infrared (H andK bands). It was developed primarily to demonstratethe utility of ground-based differential astrometry inthe search for planets around nearby stars, and todevelop key technologies for the Keck Interferometerand space-based missions.

Fig. 12. PTI aerial figure. PTI is located atopPalomar Mountain, US, next to the large white domeof the historic 5m Hale Telescope.

PTI was notable for being equipped with adual-star tracking system, the first of its kind,which simultaneously tracked interference fringesfrom a target star and a reference star against whichthe target was measured. This allowed to cancelsome of the atmospheric effects of astronomical see-ing and to make very high precision measurementspossible.

Aside from its role for the technical develop-ment of high precision dual-star astrometry, thePTI was used mainly for stellar diameter measure-ments, stellar masses and binary star work. The in-strument concluded operations in 2009.

6. EXISTING INTERFEROMETERS

6.1. NPOI

The NPOI (Navy Prototype Optical Interfer-ometer), located on Anderson Mesa near Flagstaff,Arizona, US (see Fig. 13) operates since 1995 andcombines an imaging array and an astrometric fa-cility (Armstrong et al. 1998). It can observe in

9

S. JANKOV

the visible with 32 spectral channels covering thewavelength range from 450nm to 850nm. The imag-ing subarray consists of six movable siderostats withbaseline lengths from 2.0m to 437m. The array ge-ometry has been optimized for baseline bootstrap-ping to facilitate the imaging of stellar surface struc-ture. Like the Mark III, the NPOI uses vacuum delaylines for pathlength compensation. The four-elementastrometric subarray of the NPOI includes an exten-sive site metrology system that monitors the motionsof the siderostats with respect to one another in orderto perform wide-angle astrometry with more than2mas precision.

Fig. 13. The NPOI (Navy Prototype Optical Inter-ferometer), located on Anderson Mesa, US.

6.2. SUSI

The SUSI (Sydney University Stellar Inter-ferometer) is an array of 13 telescopes with 14cmapertures, operating since 1991 at Sydney UniversityNarrabri, Australia (Davis et al. 1999, see Fig. 14).The SUSI has two beam-combining systems: Theoriginal blue system was designed to operate inthe wavelength range 400-540 nm and employs nar-row bandwidths (typically a few nanometers). Thenewer red system is designed to work in the range500-950 nm. SUSI makes observations on a singlebaseline selected from a set of fixed north-south base-lines with lengths ranging from 5m to 640m and itcan achieve angular resolutions from 20mas to 50as.

Fig. 14. The SUSI (Sydney University StellarInterferometer), Narrabri, Australia.

The 640m baseline length, the longest of allinstruments currently operational or under construc-tion, has been chosen to resolve a sample of O stars ata wavelength of 450 nm. The focus of SUSIs obser-vations is on improving our understanding of stellarastrophysics including: single stars (measuring ef-fective temperatures, radii and luminosities), binarystars, (as for single stars, plus measuring distancesand masses), variable stars (e.g. Cepheids and Mi-ras), and emission line stars (e.g. Be and Wolf-Rayetstars).

6.3. ISI

The ISI (Infrared Spatial Interferometer),University of California at Berkeley, US is locatedclose to the CHARA array and the former site ofthe Mark III on Mt. Wilson. It started operation in1990 as two-telescope interferometer but presentlyconsists of three 1.65m telescopes observing in themid-infrared. The telescopes are fully mobile andtheir current site on Mount Wilson allows for place-ments as far as 70m apart providing a resolution of3mas at 11m. On July 2003, the ISI recorded itsfirst closure phase aperture synthesis measurements.

The Fig. 15 shows three ISIs telescopes all ina line which were used for initial testing purposes,with 4m, 8m, and 12m baselines. However, they canbe moved in such a way that they form a triangleand three baselines at three different angles can bemeasured simultaneously providing the closure phasemeasurments.

Fig. 15. The ISI (Infrared Spatial Interferometer),Mt. Wilson, US.

The interferometer operates at wavelengthsbetween 9m and 12m and the stellar radiation ismixed with the output of a CO2 laser, which acts asthe local oscillator. Observations have been carriedout with baseline lengths up to 56m. One of manyadvantages of the ISIs narrow heterodyne detectionbandwidth is that one can tune the detection wave-length to be deliberately in or out of known spectrallines, allowing for interferometry on spectral lines tobe carried out.

Interest of the ISI science has mainly been fo-cused on the study of evolved stars and dust shells.However, an interferometer like the ISI is well suited

10

OPTICAL INTERFEROMETRY I

for making very precise measurements of positionsof stars and particularly to measure positions of in-frared stars which are hardly detectable in visiblelight. Investigations in the field of astrometry shouldhelp to tie the astronomical reference frames togetherwith that established in the radio, infrared and visi-ble.

6.4. CHARA

The CHARA array, named after Center forHigh Angular Resolution Astronomy, Georgia StateUniversity is located on Mt Wilson, US (see Fig.16). It consists of 6 telescopes of 1m in diameterarranged in a Y-shaped configuration with baselinesranging from 30m to 330m. First fringes on a sin-gle baseline have been obtained in November 1999and commissioning of the full array continues. Lightfrom the individual telescopes is conveyed throughvacuum tubes to the central Beam Synthesis Facil-ity in which the six beams can be combined together.When the paths of the individual beams are matchedto an accuracy of less than one micron, after the lighttraverses distances of hundreds of meters, the Arraythen acts like a single coherent telescope for the pur-poses of achieving an exceptionally high angular res-olution. The Array is capable of resolving details assmall as 200as.

Fig. 16. The facilities of the CHARA (Centerfor High Angular Resolution Astronomy) Array to-gether with the existing facilities of historic MountWilson Observatory. The Beam Synthesis Facility,an L-shaped structure of the length of a football field,can be seen close to the 2.5m telescope dome.

CHARA has entered into several collabora-tions with groups offering unique instruments ortechnologies for enhanced performance. These in-ternational collaborations have brought a significantadded value to the science capabilities of the CHARAArray. At present these collaborations include ajoint observing collaboration with the Observatoirede Paris through the FLUOR instrument which hasbeen moved from the IOTA interferometer (Section5.4) and upgraded for the CHARA Beam Synthe-sis Facility. A collaboration with the University of

Michigan has led to the development of an imagingbeam combiner (MIRC) which has already producedthe first images of stellar surfaces and close binarystars. A joint project with the University of Syd-ney has led to the development of a second beamcombiner with significantly improved sensitivity tofainter objects while also providing measurements ofvery high precision. Finally, an agreement with Ob-servatoire de la Cote dAzur has brought the thirdnew beam combiner (VEGA) to the Array capableof providing spectroscopic and polarimetric channelsfor high resolution work.

The first four-telescope fringes with VEGABeam Combiner at the CHARA Array were obtainedwith CHARA (MIRC used as the infrared fringetracker) in October 2010.

6.5. KI

The KI (Keck Interferometer), NASA JPL, onMauna Kea, Hawai, US (Fig. 17) which consists oftwo 10m Keck telescopes obtained first fringes withfull-apertures on March 2001.

Fig. 17. Two 10m telescopes of the KI (Keck In-terferometer) on Mauna Kea, Hawai, US.

The 10m telescopes are equipped with high-order adaptive optics systems, which provide goodcorrection in the near-IR and excellent wavefrontquality at 10m. It enables the implementation ofa nulling beam combiner, which will be used to char-acterize exo-zodiacal emission around nearby main-sequence stars. The standard science modes of theinstrument are:

1. V2 mode, high sensitivity visibility ampli-tude measurements in the near infrared (H and K).

2. Nulling interferometry in N-band (8 -12m) to suppress light from central star and toreveal faint extended emission around it. Key sci-entific goals are to characterize exo-zodiacal emis-sion around Sun-like stars to inform Terestial PlanetFinder (see Section 7.3) mission design.

3. Differential Phase, multi-color fringe phasemeasurements between 2-5m, with direct HotJupiter detection (orbital parameters and masses),as the key scientific goal.

11

S. JANKOV

Recently, the Keck interferometer was up-graded to do self-phase-referencing (SPR) assistedK-band spectroscopy at R 2000. This SPR modeas currently being implemented as the ASTrometricand phase-Referenced Astronomy (ASTRA) projectwill provide phase referencing and astrometric ob-servations at the Keck Interferometer, leading to en-hanced sensitivity, and the ability to monitor orbitsat an accuracy level of 30-100as, high spectral res-olution observation of Young Stellar Objects (YSO),sensitive Active Galactic Nuclei (AGN) observations,and astrometric (known exo-planetary systems char-acterization and galactic center general relativity)capabilities. With the first high spectral resolutionmode now offered to the community, this contribu-tion focuses on the progress of the dual field andastrometric modes.

It is planned that the KI will be upgradedwith four new 1.8m outrigger telescopes in order tomake possible imaging mode. A large fraction of theobserving time available with the outriggers shouldbe devoted to an astrometric search for extrasolarplanets while the combination of all six telescopesshould result in a sensitive imaging array.

6.6. VLTI

The VLTI (Very Large Telescope Interferome-ter), European Southern Observatory Paranal, Chile,obtained first fringes on the sky in March 2001, withsiderostats, and presently it allows the interferomet-ric combination of the four 8.2m telescopes (Fig. 18),augmented by four movable 1.8m auxiliary telescopes(Fig. 19).

The first generation of instruments included acommissioning instrument based on a two-telescopenear-IR beam combiner VINCI (now decommis-sioned) as well as actually operating instrumentsMIDI, AMBER and PRIMA:

MIDI (MID-infrared Interferometric instru-ment) is the mid-infrared (N-band = 8 to 13m)combiner which combines two beams, either from theVLT main 8.2m Unit Telescopes (UTs) or from the

Fig. 18. Four 8.2m main Unit Telescopes ofthe VLTI (Very Large Telescope Interferometer),Paranal, Chile.

Fig. 19. Four 1.8m Auxilary Telescopes ofthe VLTI (Very Large Telescope Interferometer),Paranal, Chile.

1.8m Auxiliary Telescopes (ATs), to provide visibil-ity moduli in the (u, v) plane. MIDI features spectro-scopic optics to provide visibilities at different wave-lengths within the N band and with spectral resolu-tion R=30 with the prism, R=230 with the grism (acombination of a diffraction grating and prism). UTadaptive optics guarantees diffraction-limited imagequality on MIDI, for targets brighter than V = 17m.These unique capabilities allowed an access to vari-ous observing programs including observation of theGalactic Center, analysis of the nuclei of the AGNgalaxies, particularly stellar and circumstellar ob-jects, as well as solar system objects.

AMBER is the near-infrared instrument forthe VLTI, which offers the possibility of combiningtwo or three beams from either the UTs or the ATs.With spectral resolution up to 10 000, high visibilityaccuracy and the ability to obtain closure phases,AMBER offers the means to perform high qualityinterferometric measurements in the J,H,K bands (1-2.5m range). Angular resolution is set by the max-imum available baseline, which is about 200 metersfor the ATs and about 130 meters for the UTs. Ac-cordingly, the limit is about 2mas for the ATs, andabout 3mas for the UTs, in the K band. These de-sign characteristics, coupled to the VLT interferome-ter potential, open up the access to investigations ofseveral classes of objects, from stellar to extragalac-tic astronomy including, for instance, the solar sys-tem objects, low-mass stars, Cepheids, microlensing,hot exo-planets, the galactic center, nearby galax-ies, and Young Stellar Objects, Luminous Blue Vari-ables, Be stars, Novae, Wolf-Rayet and Post-AGBstars for which the results have been already pub-lished.

PRIMA (Phase-Referenced Imaging andMicro-arcsecond Astrometry) is a system designedto enable simultaneous interferometric observationsof two objects that are separated by up to 1 ar-cmin, without requiring a large continuous field ofview. PRIMA improves the sensitivity by using twoobjects (a bright star guide near a fainter targetobject) to bring the corrections to the atmosphericturbulences. Depending on the operation mode,

12

OPTICAL INTERFEROMETRY I

PRIMA can be used either to measure the angu-lar separation between the two objects (astrometrymode) or to produce images of the fainter of the twoobjects using a phase reference technique (imagingmode). With the limiting magnitude K 19m withthe UTs and K 15m with the ATs and the angulardistance between the targets that can be evaluatedwith a resolution of 10as, the objects such as extra-solar planets, galactic center, Active Galactic Nucleiand extragalactic objects as well as gravitationalmicro-lensing are the targets for PRIMA.

In addition to these ESOs instruments thereis a possibility to bring and install the visiting instru-ments. One of the visiting instruments PIONIER,developed at LAOG in Grenoble, complements ex-isting AMBER and MIDI instruments. While AM-BER has previously combined the light from threeof the telescopes at the VLTI, the light coming fromthe four 1.8-metre Auxiliary Telescopes has been suc-cessfully combined for the first time, in October 2010.This is an important step towards unleashing the fullpotential of the VLTI to use multiple telescopes to-gether.

The second generation of VLTI instrumentsGRAVITY, VSI and MATISSE are under develop-ment:

GRAVITY will coherently combine the lightfrom all four UTs. It will provide near infrared adap-tive optics assisted precision narrow-angle astrom-etry (10as) and interferometric phase referencedimaging (4mas) of faint objects (K=20m). It willprecisely measure the fringe visibility and phase ofthe fringe pattern to acquire spatial knowledge of theobserved object. With an accuracy of 10as, GRAV-ITY will be able to study motions to within a fewtimes the event horizon size of the Galactic Centermassive black hole and potentially test the GeneralRelativity in its strong field limit. It will be able todetect intermediate mass black holes throughout theGalaxy by their gravitational action on surroundingstars. It will directly determine masses of exo-planetsand brown dwarfs, and trace the origin of protostel-lar jets. Also it will be capable of spatially resolvingcoherent gas motions in the broad line regions of Ac-tive Galactic Nuclei in external galaxies. Through itshigh performance infrared wavefront sensing system,GRAVITY will open up deep interferometric imag-ing studies of stellar and gas components in dusty,obscured regions, such as obscured active galacticnuclei, dust-embedded star forming regions, and pro-toplanetary disks.

VSI (V(LTI) Spectro-Imager) was proposedas a second-generation instrument in order to pro-vide the community with spectrally-resolved, near-infrared images at angular resolutions down to1.1mas and spectral resolutions up to R = 12 000.Targets as faint as K = 13m will be imaged withoutrequiring a brighter nearby reference object whilefainter targets can be accessed if a suitable refer-ence is available. The unique combination of high-dynamic-range imaging at high angular resolutionand high spectral resolution enables a scientific pro-gram which serves a broad user community and atthe same time provides the opportunity for break-throughs in many areas of astrophysics. The highlevel specifications of the instrument are derived from

a detailed science case based on the capability toobtain milli-arcsecond resolution images of a widerange of targets including: probing the initial condi-tions for planet formation in the AU-scale environ-ments of young stars; imaging convective cells andother phenomena on the surfaces of stars; mappingthe chemical and physical environments of evolvedstars, stellar remnants, and stellar winds, and disen-tangling the central regions of active galactic nucleiand supermassive black holes. The VSI will providethese new capabilities using technologies which havebeen extensively tested in the past. It is quite flexi-ble instrument and should be able to make maximumuse of new infrastructure as it becomes available; forexample, by combining up to 8 telescopes, enablingrapid imaging through the measurement of up to 28visibilities in every wavelength channel within a fewminutes. The current studies are focused on a 4-telescope version with an upgrade to a 6-telescopeone.

MATISSE (Multi AperTure mid-InfraredSpectroScopic Experiment) is foreseen as a mid-infrared spectro-interferometer combining up to fourUTs or ATs beams. It will measure closure phase re-lations and thus offer an efficient capability for imagereconstruction. In addition to the N band, the MA-TISSE will open three new observing windows at theVLTI: the L, M, and Q bands, all belonging to themid-infrared domain. Furthermore, the instrumentwill offer the possibility to perform simultaneous ob-servations in separate bands while providing severalspectroscopic modes. The unique performance of theinstrument is related to the existence of the four UTsthat permits to push the sensitivity limits to valuesrequired by selected astrophysical programs such asthe study of Active Galactic Nuclei and protoplane-tary discs. Moreover, the evaluated performance ofMATISSE is linked to the availability of ATs whichare relocatable in position in about 30 different sta-tions allowing the exploration of the Fourier planewith up to 200 meters baseline length. Key scienceprograms using the ATs cover for example the forma-tion and evolution of planetary systems, the birth ofmassive stars as well as the observation of the high-contrast environment of hot and evolved stars. Insummary, MATISSE can be seen as a successor ofMIDI by providing imaging capabilities in the en-tire mid-infrared accessible from the ground. Theextension of MATISSE down to 2.7m as well as itsgeneralization of the use of closure phases makes italso a successor of AMBER.

6.7. LBTI

The LBTI (Large Binocular Telescope Inter-ferometer) on Mt. Graham, Arizona, US (Fig. 20)consists of two 8.4m telescopes mounted side by sidein a single alt-azimuth mount, with a 14.4m center-to-center spacing.

This configuration offers some unique capabil-ities for interferometry, as it lends itself to Fizeaubeam combination with a wide field of view andlow thermal background. It is a fully cryogenic in-strument devoted to near and mid-infrared observa-tions (2-20m). Among the unique characteristics ofthe telescope are the two adaptive secondary mirrorswhich enable diffraction limited seeing. The system

13

S. JANKOV

Fig. 20. The LBT (Large Binocular Telescope) onMt. Graham, US.

is designed for high spatial resolution, high dynamicrange imaging and nulling interferometry. It con-sists of an universal beam combiner (UBC) and threecamera ports, one of which is populated currently bythe Nulling and Imaging Camera (NIC).

One of the key projects for the LBTI is to sur-vey nearby stars for debris disks down to levels whichmay obscure detection of Earth-like planets. A sur-vey of approximately 80 nearby stars that is essentialas a ground-based precursor of the space missions (asfor example TPF/Darwin, see Sections 7.3 and 7.4).A survey will identify zodiacal disks, which signal theexistence of a planetary system for follow up studywith TPF/Darwin. On the other hand, the surveywill also weed out those systems with very intense zo-diacal emission, which can overwhelm the signal ofterrestrial planets. Finally, the survey can also ad-dress the question of zodiacal emission strength as afunction of spectral type and age of the central star.

In addition, the consortium of German insti-tutes and observatories led by the Max Planck Insti-tute for Astronomy in Heidelberg together with theIstituto Nazionale di Astrofisica in Arcetri, is devel-oping the LINC ((L)BT (IN)terferometric (C)amera)instrument. The LINC will combine the radiationfrom the two 8.4m primary mirrors in Fizeau modewhich allow true imagery over a wide field of view.The beam combiner will operate at wavelengths be-tween 0.6 and 2.4m. When coupled with the ad-vanced adaptive optics system of the LBT, the LINCinstrument will allow the interferometric measure-ments over a field of approximately 10-20 arcsecondssquare. Ultimately, after improvement of the Adap-tive Optics, which will allow better performance overa wider field of view, the instrument will be knownas NIRVANA, the (N)ear-(IR) / (V)isible (A)daptiveI(N)terferometer for (A)stronomy. The main scien-tific goals of LINC/NIRVANA are YSOs environ-ments, circumstellar disks and outflows, formationof binary stars, stellar clusters, the compact H IIregions, astrometry for extra-solar planets, the so-lar system minor bodies, the galactic center and thegalaxies at z 1 2.

In October 2010 the LBTI acquired its firstfringes on the sky and will be interferometricallyfunctional in 2011 once the adaptive optics and theLBTI are commissioned.

6.8. MROI

The MROI (Magdalena Ridge Observatory In-terferometer) is presently under construction andshould be operational in years to come. It is an inter-national scientific collaboration between New MexicoTech (US) and the University of Cambridge (UK).The Interferometer Projects mission is to developa ten-element imaging interferometer to operate atwavelengths between 0.6m and 2.4 m with base-lines from 7.5m to 340m. Its primarily technical andscientific goals are to produce model-independent im-ages of faint and complex astronomical targets atresolutions over 100 times that of the Hubble SpaceTelescope in order to study: star and planet for-mation, stellar accretion and mass loss, and activegalactic nuclei.

The 2.4-meter telescope achieved first light onOctober 2006 and commenced operations on Septem-ber 2008. The work at the interferometer site startedin September 2006 and the Beam Combining Facil-ity was completed in early 2008. Presently, the tele-scopes are nearing their first factory acceptance tests.When finished, the interferometer will be upgradedwith ten telescopes, each approximately 1.4 metersin diameter. The telescopes making up the interfer-ometer will be spaced by distances up to 400 metersand will be optically linked in order to make aper-ture synthesis imaging. This setup will simulate theresolving power of a single telescope up to 400 me-ters in diameter. As a result of the large numberof telescopes in the array, the interferometer will beable to make accurate images of complex astronom-ical objects.

7. FUTURE INTERFEROMETERS

7.1. OHANA

The OHANA (Optical Hawaiian Array forNanoradian Astronomy) is the project of a hecto-metric fibered array at Mauna Kea, Hawaii. Theconcentration of large apertures on Mauna Kea (twoKeck, Gemini, Subaru, CFHT, UKIRT) lends itselfto ideas for interferometric combination of these sixtelescopes (Mariotti et al. 1996). By taking advan-tage of the existing adaptive optics systems, such asalready installed on telescopes, this would provideexcellent sensitivity and unprecedented imaging ca-pabilities with baselines ranging from 75m to 800m(the largest baseline of OHANA (Subaru-Gemini)).This yields resolutions of 0.25mas and 0.5mas at 1and 2 microns respectively, therefore opening thepossibilities for research in many fields as for VLTI,but with four times better spatial resolution.

The OHANA project has carried out initialexperiments to couple the light from Mauna Keatelescopes into a singlemode pair of 300m fibres andachieved first fringes with small prototype interfer-

14

OPTICAL INTERFEROMETRY I

ometer on July 2010, the first step in a plan to linkthe giant telescopes of Hawaii into a powerful opticalinterferometer.

7.2. Interferometry at Antartica

The Dome C at Antartica (altitude 3300m) isconsidered as a perfect site for astronomical inter-ferometry due to low wind speeds, little atmosphericturbulence, negligible seismic activity and very sta-ble climate. For this reason, several interferometricprojects are under development, most notably theAPI (Antartic Planet Interferometer, Swain et al.2003) and KEOPS (Kiloparsec Explorer for OpticalPlanet Search, Vakili et al. 2003).

The API should be built in two phases. Phase1 is proposed to consist of a two-element active fringetracking interferometer using two 50cm diametertelescopes operating in a direct visibility (V2) modewith baselines of up to 400 metres. The H,K,L,Mbands (1.4 to 5.4m) are to be targeted to give max-imum resolution and sensitivity at up to 104 spectralresolution. In Phase 2, the full API, will consist ofseveral 2m diameter telescopes and will be capable ofoperating in a variety of modes. It will be a partic-ularly powerful tool for the study of extrasolar plan-ets, protoplanets, Young Stellar Objects, and ActiveGalactic Nuclei accretion disks.

The KEOPS is proposed as three concentricrings of 1 to 2m diameter telescopes covering a di-ameter of 650 metres. The 36 co-phased primarytelescopes would have an equivalent collecting areaof at least 60m2 and a resolution of a milli-arcsecondat 10m. KEOPS is designed for direct imaging ofexo-planets in the thermal infrared, as well as forastroseismology studies.

7.3. TPF

The TPF (Terrestrial Planet Finder) Interfer-ometer is a concept for a formation flying interferom-eter to search for Earth-like planets around nearbystars, planet studies and signs of life. TPF was be-ing developed as a possible collaboration with ESAsDarwin mission but the project was recently can-celled.

The NASAs 2010 Decadal Survey recom-mended a continuation of technology developmentfor nulling interferometry under the title of NewWorlds Technology Development Program and now,although TPF no longer exists as a planned mis-sion within NASA, some technology work continues.Efforts in 2009 and 2010 have included broadbandnulling with the Adaptive Nuller and demonstrationswith the Planet Detection Testbed. One of the pos-sible configurations of a space interferometer (an op-tion considered for Terrestial Planet Finder) is pre-sented in Fig. 21.

At present, NASA continues to work with Uni-versity scientists and industry to develop the TPFtechnology. Meanwhile, the Kepler mission will col-lect information on the number of possible nearbyEarthlike planets, while the Keck Interferometer andLarge Binocular Telescope Interferometer projectswill measure, for this purpose, the dust surroundingthose stars.

Fig. 21. A possible configuration of a space inter-ferometer.

7.4. Darwin

The Darwin mission was planned as a con-stellation of four or five spacecrafts, with primarygoal to look for Earth-like planets and analyse theiratmospheres for chemical signatures of life. Onespacecraft would have acted as a central communica-tions station. The other three/four would have func-tioned as light collectors, redirecting light beams tothe central station. The constellation was also in-tended to carry out high-resolution imaging usingaperture synthesis, to provide images of celestial ob-jects with unprecedented detail. Instead of orbit-ing Earth, Darwin was proposed to be placed be-yond the Moon. At a distance of 1.5 million kmfrom Earth, in a direction opposite to that of theSun, Darwin would have operated from the secondLagrange point L2. The Darwin mission was pro-posed in 2007 as a concept for Phase A Study withinESAs Cosmic Vision, which includes missions be-yond 2015. It was not chosen for further study, but asimilar project should be re-proposed to ESA. What-ever the final design of the telescopes, it is clear thatthe TPF/Darwin technology will be a pathfinder forfuture micro and nano-arcsecond resolution instru-ments at infrared and other wavelengths.

8. CONCLUSION

After decades of development, optical interfer-ometry is now beginning to play a major role in astro-physics. The emergence of large interferometers, inparticular the Very Large Telescope interferometer,CHARA, the Keck Interferometer and Large Binocu-lar Telescope Interferometer promise to revolutionizethe impact of high resolution observations in manyareas of astrophysics. Current programs execute in-creasingly varied and sophisticated observations. Ini-tially, these were almost exclusively observations ofstars and circumstellar phenomena, but recently, op-tical interferometry extended even to extragalactictargets.

The field is currently driven forward by the ac-tivities in many research areas, and important scien-tific results are expected in the near future. Clearly,the main beneficiaries will be stellar astrophysics and

15

S. JANKOV

galactic astronomy, in particular the areas of fun-damental stellar properties, star and planet forma-tion, and all stages of stellar evolution, but increas-ingly, optical interferometry will play more impor-tant role in other areas of astrophysics particularlyexo-planetary and extragalactic research.

Acknowledgements This work has been supportedby the Ministry of Science and Technological Devel-opment of the Republic of Serbia (Project No 146007Inverse problems in astrophysics : Interferometryand Spectrophotometry of stars).

REFERENCES

Anderson, J. A.: 1920, Astrophys. J., 51, 263.Armstrong, J. T., Mozurkewich, D., Rickard, L. J.,

Hutter, D. J., Benson, J. A. et al.: 1998, As-trophys. J., 496, 550.

Assus, P., Choplin, H., Corteggiani, J. P., Cuot, E.,Gay, J., Journet, A., Merlin, G. and Rabbia,Y.: 1979, J. Opt., 10, 345.

Baldwin, J. E., Beckett, M. G., Boysen, R. C. et al.:1996, Astron. Astrophys., 306, L13.

Beckers, J. M.: 1982, Optica Acta, 29, 361.Blazit, A., Bonneau, D., Josse, M., Koechlin, L.,

Labeyrie, A. and Oneto, J. L.: 1977, Astro-phys. J., 217, L55.

Boccaletti, A., Riand, P., Moutou, C., Labeyrie, A.:2000, Icarus, 145, 628.

Bracewell, R. N.: 1978, Nature, 274, 780.Colavita, M. M., Wallace, J. K., Hines, B. E., Gursel,

Y., Malbet, F. et al.: 1999, Astrophys. J.,510, 505.

Cornwell, T. J.: 1987, Astron. Astrophys., 180, 269.Coude Du Foresto, V., Ridgway, S., Mariotti, J. M.:

1997, Astron. Astrophys. Suppl. Series, 121,379.

Davis, J., Morton, D. C., Allen, L. R. and HanburyBrown, R.: 1970, Mon. Not. R. Astron. Soc.,150, 45.

Davis, J., Tango, W. J., Booth, A. J., Brummelaar,T. A., Minard, R. A., Owens, S. M.: 1999,Mon. Not. R. Astron. Soc., 303, 773.

Di Benedetto, G. P., Conti, G.: 1983, Astrophys. J.,268, 309.

Di Benedetto, G. P.: 1985, Astron. Astrophys., 148,169.

Di Benedetto, G. P., Rabbia, Y.: 1987, Astron. As-trophys., 188, 114.

Fizeau, H.: 1868, C. R. Acad. Sci., 66, 932.Gay, J., Journet, A.: 1973, Nature Phys. Sci., 241,

32.Hanbury Brown, R., Twiss, R.: 1956a, Nature, 177,

27.Hanbury Brown, R., Twiss, R.: 1956b, Nature, 178,

1046.Hanbury Brown, R., Davis, J., Allen, L. R.: 1967a,

Mon. Not. R. Astron. Soc., 137, 375.Hanbury Brown, R., Davis, J., Allen, L. R., Rome,

J. M.: 1967b, Mon. Not. R. Astron. Soc.,137, 393.

Hanbury Brown, R., Davis, J., Herbison-Evans, D.,Allen, L. R.: 1970, Mon. Not. R. Astron.Soc., 148, 103.

Hanbury Brown, R., Davis, J., Allen, L. R.: 1974a,Mon. Not. R. Astron. Soc., 167, 121.

Hanbury Brown, R., Davis, J., Lake, R. J. W.,

Thompson, R. J.: 1974b, Mon. Not. R. As-tron. Soc., 167, 475.

Jankov, S., Vakili, F. Dominiciano de Souza, Jr.,Janot-Pacheco, E.: 2001, Astron. Astrophys.,377, 721.

Jankov, S., Vakili, F., Domiciano de Souza Jr., A.,Janot-Pacheco, E.: 2002, ASP Conf. Se-ries, eds. C. Aerts, T. R. Bedding and J.Christensen-Dalsgaard, 259, 172.

Jankov, S., Vakili, F., Domiciano de Souza, A.: 2003IAU Symp., 210, 16.

Jennison, R. C.: 1958, Mon. Not. R. Astron. Soc.,118, 276.

Knox, K. T., Thompson, B. J.: 1974, Astrophys. J.,193, L45.

Labeyrie, A.: 1970, Astron. Astrophys., 6, 85.Labeyrie, A.: 1975, Astrophys. J., 196, L71.Labeyrie, A.: 1996, Astron. Astrophys. Suppl. Se-

ries, 118, 517.Mariotti, J. M., Coude du Foresto, V., Perrin, G.,

Zhao, P., Lena, P.: 1996, Astron. Astrophys.Suppl. Series, 116, 381.

Michelson, A. A.: 1891a, Publ. Astron. Soc. Pacific,3, 274.

Michelson, A. A.: 1891b, Nature, 45, 160.Michelson, A. A.; Pease, F. G.: 1921, Astrophys. J.,

53, 249.Mourard, D., Tallon-Bosc, I., Blazit, A., Bonneau,

D., Merlin, G. et al.: 1994, Astron. Astro-phys., 283, 705.

Mourard, D., Clausse, J. M., Marcotto, A., Perraut,K., Tallon-Bosc, I., Brio, Ph., Blazit, A., Bon-neau, D., Bosio, S., Bresson, Y. et al.: 2009,Astron. Astrophys., 508, 1073.

Petrov, R. G.: 1988, Diffraction-Limited Imagingwith Very Large Telescopes, eds. D. M. Al-loin, J. M. Mariotti (Kluwer), 249.

Roddier, F.: 1986, Opt. Commun., 60, 145.Ryle, M., Vonberg, D. D.: 1946, Nature, 158, 339.Schwarzschild, K.: 1896, Astron. Nachr., 139, 353.Shao, M., Staelin, D. H.: 1980, Appl. Opt., 19, 1519.Shao, M., Colavita, M. M., Hines, B. E., Staelin, D.

H., Hutter, D. J.: 1988, Astron. Astrophys.,193, 357.

Shao, M., Colavita, M. M.: 1992, Astron. Astro-phys., 262, 353.

Stephan, E.: 1874, C. R. Acad. Sci., 78, 1008.Swain, M. R., Coude du Foresto, V., Fossat, E., Vak-

ili, F.: 2003, Mem. Soc. Astron. It. Suppl.,2, 207.

Tango, W. J.: 1990, Appl. Opt., 29, 516.Thom, C., Granes, P., Vakili, F.: 1986, Astron. As-

trophys., 165, L13.Tuthill, P. G., Monnier, J. D., Danchi, W. C., Wish-

now, E. H., Haniff, C. A.: 2000, Publ. Astron.Soc. Pacific, 112, 555.

Vakili, F., Aristidi, E., Fossat, E., Abe,L., Domi-ciano, A., Belu, A., Agabi, A., Schmider,F. X., Lopez, B., Swain, M.: 2003, SF2A-2003: Semaine de lAstrophysique Francaise,eds. F. Combes, D. Barret and T. Contini,EdP-Sciences, 365.

Vakili, F., Aristidi, E., Abe, L., Lopez, B.: 2004,Astron. Astrophys., 421, 147.

Weigelt, G. P.: 1977, Opt. Commun., 21, 55.Young, J. S. et al.: 2000, Mon. Not. R. Astron.

Soc., 315, 635.

16

OPTICAL INTERFEROMETRY I

ASTRONOMSKA OPTIQKA INTERFEROMETRIJA.I METODE I INSTRUMENTI

S. Jankov

Astronomical Observatory Belgrade, Volgina 7, 11060 Belgrade 38, SerbiaEmail: sjankov@aob.rs

UDK 520.3613Pregledni rad po pozivu

Prethodna dekada je videla ostva-ree velikih interferometrijskih projeka-ta ukuqujui teleskope klase 8-10m istometarske interferometrijske baze. Mo-derni kompjuteri i kontrolna tehnologijasu omoguili interferometrijsku kombi-naciju svetlosti sa razdvojenih teleskopakako u vidivom tako i u infracrvenomreimu. Oslikavae sa rezolucijom od mili-luqne sekunde i astrometrija sa preciznoxuod mikro-luqne sekunde su tako postale real-nost. Ovde dajem pregled metoda i instru-mentacije koje odgovaraju tekuem stau uoblasti astronomske optiqke interferomet-rije. Prvo, ovaj pregled sumarizuje razvoj odpionirskih radova Fizoa i Majkelsona. Za-tim su opisane osnovne dostupne veliqine aposle toga sledi diskusija o osnovnim prin-

cipima konstrukcije modernih interferome-tara. Osnovne interferometrijske tehnike kaoxto su interferometrija spekl i aper-turnih maski, sinteza aperture i nu-ling interferometrija su takoe diskuto-vani. Koristei iskustvo steqeno od prox-lih i postojeih instrumenata da bih ilus-trovao vane stavke, posebno razmatram novugeneraciju velikih interferometara koji sunedavno staveni u upotrebu (posebno QARA,KEK, VLT i LBT interferometri). Na krajudiskutujem dugoroqnu budunost optiqke in-terferometrije ukuqujui mogunost novihvelikih projekata za zemaske opservatorijekao i predviaa za kosmiqku interferomet-riju.

17