Towards the Intensity Interferometry Stellar Imaging System
M. Daniel, W.J. de Wit, D. Dravins, D. Kieda, S. LeBohec, P. Nunez, E. Ribak
aa r X i v : . [ a s t r o - ph . I M ] J un Towards the Intensity Interferometry Stellar Imaging System
M. Daniel, W.J. de Wit, D. Dravins, D. Kieda,S. LeBohec, P. Nunez, E. Ribakfor the Stellar Intensity Interferometry working group(within IAU commission 54)October 25, 2018
Contact : David Kieda, University of Utah, Phone: 801 581 6901,Email: [email protected]
Abstract
The imminent availability of large arrays of large light collectors deployed to exploit at-mospheric Cherenkov radiation for gamma-ray astronomy at more than 100 GeV, motivatesthe growing interest in application of intensity interferometry in astronomy. Indeed, plannedarrays numbering up to one hundred telescopes will offer close to 5,000 baselines, rangingfrom less than 50 m to more than 1000 m. Recent and continuing signal processing technologydevelopments reinforce this interest. Revisiting Stellar Intensity Interferometry for imaging iswell motivated scientifically. It will fill the short wavelength (B/V bands) and high angularresolution ( < . Key science goals
Stellar astrophysics distinguishes itself by the great progress that has been made in understandingthe origin and evolution of the basic astrophysical phenomena, despite a physical inability to directlyimage the stellar surfaces. Nearby stars have maximum angular extent of tens of milli-arcseconds(mas); direct imaging of objects in this angular range requires use of optical interferometry withbaselines of one hundred meters or more. Until now, only a handful of stars (in addition to theSun) have been directly imaged to some extent using interferometric techniques.Of numerous interferometric methods, two are especially well suited for imaging of stellar sur-faces. Stellar amplitude (Michelson) interferometry (SAI) was employed (very simply) in the 1920’sto measure a few cool stars, but the technique was strongly control-technology-limited. In the 1960’s,stellar intensity interferometry (SII) was employed to measure some 32 stars [13], but it was, atthat time, very sensitivity-limited. More recently, SAI has been extensively developed, with sev-eral arrays offering telescope apertures in the range 0.5-10 meters and baselines up to ∼
300 meters.Modern amplitude interferometry has achieved sub-mas resolution, with many applications to stellarphysics Amplitude interferometry as currently deployed, however, has significant limitations. In orderto image hot stars (typical diameters < area. Such dense sam-pling of an optical signal over the large area can sample the interferometric u-v plane with close to5,000 baselines ranging from a few tens of meters to more than a kilometer. The combination of thedense sampling as well as the shorter wavelength measurements can allow the u-v interferometricplane coverage to potentially measure stellar diameters down to the 0.1 mas domain, a capabilitythat will open up a new field of ultra-high precision ground-based astronomical imaging. SII wouldoffer a very complementary functionality to a Cherenkov array, as the SII operation, on relativelybright stars, would be far less sensitive to sky brightness, and could operate under moonlight, duringtimes not useful for Cherenkov studies.We envision a two-stage development and implementation of modern SII. In the first stage, SIIcould be implemented on one or more existing Cherenkov arrays, immediately offering the advan-tages of blue operation and high resolution. In a second stage, SII could be implemented as anaugmentation of a next generation Cherenkov array, with an extremely rich range of baselines pro-viding very high resolution and imaging capability unlikely to be matched by SAI in the foreseeablefuture, and in particular opening the possibility of very detailed imaging of young stars.The ongoing developments in this direction are presented in the following sections after anoutline of the science potential offered by the proposed re-deployment of SII. The capability toaddress various questions depends on the actual design of a modern intensity interferometer. Inthe following sub-sections, in order to discuss the value of a few selected science topics which could See the numerous Astro2010 Science White papers http://usic.wikispaces.com/ASTRO2010+White+Papers
2e advantageously investigated, we adopt a conservative limiting visual magnitude m v < Pre-main sequence stars are young stars that are contracting towards the main-sequence, but arepowered only by gravitational collapse: core temperature and pressure are insufficient to ignitehydrogen fusion. Key questions relating to the physics of mass accretion and PMS evolution canbe addressed by means of very high resolution imaging. Spatially resolved studies will address theabsolute calibration of PMS tracks, the mass accretion process, continuum emission variability, andstellar magnetic activity, including the formation of potential jet emission.Features on the stellar surface may also be directly imaged. Hot spots on the stellar surfacedeliver direct information regarding rotation via the von Zeippel effect, as well as tracing theaccretion of material onto the stellar surface. Cool spots, on the other hand, may cover 50% ofthe stellar surface; they may be produced by magnetic field effects (sun-spots), or as a productof the slowly decaying rapid rotation of young stars. Measurement of the number, distribution,and then variations of hot and cool spots will constrain the interplay of rotation, convection, andchromospheric activity. It may also provide direct practical application for exploring the anomalousphotometry observed in young stars[34]. In addition, the presence of hot and cold features on thestellar surface is not only limited to PMS stars; observation of such features on main-sequencestars may allow detailed study of sun-like transient phenomena on a wider class of main sequencestars. Because the most luminous stars will provide the strongest interferometric signal, U/V bandmeasurements of stellar surface features by SII can provide optimal contrast.In the last decade, several young, coeval stellar groups have been discovered in close proximity( ∼
50 pc) to the sun. Prominent examples of nearby coeval stellar groups include the TW Hydraand β Pic co-moving groups. The proximity of the co-moving groups ensures that their membersare bright. The majority of the spectral types within these nearby groups range between A andG-type, and about 50 young stars have m v <
8. Their close proximity renders the co-movinggroups relatively sparse making them very suitable, un-confused targets. The relative sparsenessalso must certainly generate incomplete group membership. As stellar distances continue to berefined, additional members of these groups may be discovered, making it likely that the knownnumber of young, bright stars will continue to increase with time. The ages of the stellar coevalgroups lie in the range between 8 to 50 Myr (see [40] for an overview). These young ages imply thata substantial fraction of the low-mass members are still in their PMS contraction phase.The evolutionary PMS tracks in the Hertzsprung-Russell diagram are usually calibrated usingspectroscopic binaries, which do not allow to resolve the individual stellar masses. Statisticalmethods are used to estimate the inclination angle of the systems and estimate the average massesfor an ensemble of stars. By directly imaging the binary orbital parameters, the inclination of thesystem can be determined, leading to direct measurement of the masses of the binary components.This is a fundamental exercise not only for PMS stars but for all spectroscopic binaries in anyevolutionary stage. Measurement of angular sizes of individual PMS stars in combination with adistance estimate (e.g. GAIA) allows a direct comparison between predicted and observed sizes ofthese gravitationally contracting star on a star-by-star basis, rather than on an average basis overan ensemble of stars.
Measuring diameters of Cepheids is a basic method with far reaching implications for determinationof the distances of nearby galaxies. Currently, optimum use of a Cepheid distance scale requires3alibration of distances, and correction for Cepheid atmospheric factors. Hipparcos (and in thefuture GAIA and perhaps SIM) provide precision distances for calibration. SII promises muchhigher resolution and extension of Cepheid calibration to greater distances with the Baade-Wesselinkmethod [33]. In an exciting development, intensity interferometry was recently proposed as anapproach for the measurement of absolute distances by exploiting the time curvature of the observedlight wave fronts [32]. Meanwhile, SAI is already providing critical ancillary information, such asthe atmospheric p factor, for Cepheids of known distance [25].
Classical Be stars are particularly well-known for their close to break-up rotational velocities asdeduced from photospheric absorption lines. In addition, they show both IR excess and Balmerline emission due to a gaseous circum-stellar disk. The line emission appears and disappears ontimescales of months to years. Photometric observations of Be star disks sometimes provide evidencefor evolution of the disk structure into several ring structures, before the gas disappears into theinterstellar medium (e.g. [6]). The Be-phenomenon is fairly common (fraction of Be stars to normalB-type peaks at nearly 50% for B0 stars, [39]), and is therefore indicative of a fundamental stellarphysics phenomenon. There are about 300 Be stars brighter than m v = 8, roughly correspondingto a distance limit of 700 pc.The disk structure, fast rotation, and emission line phenomena in Be stars appear to be related.Although the exact underlying Be mechanism has yet to be identified, the Be phenomenon isprobably related to the high rotational velocity of the star. Absorption line studies cannot providean accurate rotational velocity in these objects due to strong gravity darkening at the equator andbrightening at the pole areas. However, direct measurement of the physical shape of the rotatingstar can provide a direct measure of the rotational speed (see α Eri with the VLTI, [5]). The peakblackbody wavelengths of the Be spectra are in the U/V bands. Since ground-based Michelsoninterferometry is still restricted to IR wavelengths for imaging, the long-baseline coverage of theinterferometric u-v plane is reduced , even for the same physical telescope baseline. In addition,the lower photon flux in the IR tail further decreases the number of observable Be stars.Since SII is substantially less sensitive to atmospheric turbulence than Michelson interferometry,observations of Be stars in the U/V band can enable high resolution studies of Be stars oversubstantially longer baselines. A rough estimate of the effective temperatures of the objects in theBright Star Catalog reveals there are approximately 2600 stars brighter than visual magnitude 7and hotter than 9000 K. Their typical angular sizes range between 0.5 mas and 5 mas making themobservable even with a fairly modest scale intensity interferometer array.
Optical interferometry is still in its infancy, but already more than 150 science publications showthe dramatic return from high angular resolution studies of stars, ranging from YSO’s and pre-planetary disks, to MS stars, debris disks, multiple stars, mass loss in all kinds of systems, andPMS evolution. SII promises to complement and extend such programs to blue/visible imagerywith much higher resolution, and at comparatively modest incremental cost. Rather than developspecific science topics further here, we refer to the typically 4-5 SAI publications currently appearingin the literature each month. Technical overview
The intensity interferometry technique does not rely on the actual interference of light rays as inMichelson interferometry. Instead, the interferometric signal, the degree of mutual coherence, ischaracterized by the degree of correlation of light intensity fluctuations observed at two differentdetectors. In practice, the degree of mutual coherence is measured using fast temporal correla-tions between narrow optical waveband intensity fluctuations observed by two (or more) telescopesseparated by a baseline distance.The principal component of the intensity fluctuation is the classical shot noise which will notdemonstrate any correlation between the two separated telescopes. The intensity interferometricsignal is related to a smaller noise component: the wave noise. The wave noise can be understoodas the ’beat frequency’ in optical intensity between the different Fourier components of the lightreaching the telescopes. This wave noise will show correlation between the two detectors, providedthere is some degree of mutual coherence between the light received at the two telescopes. As perequation 1, this intensity correlation (the time integrated product of the intensity fluctuations ∆i and ∆i in two telescopes) is positive. The correlation provides a measurement of the square ofthe degree of coherence of the light at the two detectors | γ (d) | (the fringe visibility in a Michelsoninterferometer) where d represents the distance separating the telescopes. [23]. | γ ( d ) | = < ∆ i · ∆ i >< i >< i > (1)According to the van Cittert-Zernike theorem, the complex degree of coherence ( γ (d)) is thenormalized Fourier transform of the source intensity angular distribution. In the case of a sourcethat can be modeled as a uniform disk with angular diameter θ , the various baselines used in theobservation sample an Airy function on the ground. As a consequence, when the baseline d is toosmall for resolving the observed object, | γ (d) | = 1 and ∆i and ∆i are maximally correlated.Conversely, when the telescope baseline is sufficient to resolve the Airy disk, | γ (d) | <
1, and∆i and ∆i are still correlated to some extent. For observations at wavelength λ , the degree ofcoherence γ (d) = 0 when d = 1 . λ/θ . An intensity interferometer with 1 km baseline operatingat a 400 nm wavelength should therefore be able to probe features with angular extent less than0.1 mas.The important point is that the technique relies on the correlation between the (relatively) low-frequency intensity fluctuations between different detectors, and does not rely on the relative phaseof optical waves at the different detectors (see Figure 1). The requirements for the mechanical andoptical tolerances of an intensity interferometer are therefore much less stringent than in the caseof a Michelson interferometer. One strong advantage of intensity interferometry is therefore itsmechanical and atmospheric robustness: the required accuracy depends on the electrical bandwidthof the detectors, and not on the wavelength of the light. This opto-mechanical robustness alsomeans that the atmosphere does not influence the performance of the instrument even in the U andV bands.The major drawback of SII is that copious quantities of light are necessary to observe the wavenoise signal in the presence of the much larger shot noise. Very large light collectors (50 −
100 m area) are necessary to have sufficient statistical strength to observe non-Poisson fluctuations aroundthe mean photon intensity. However, the tolerance of SII to path length differences makes such lightcollectors relatively inexpensive. An intensity interferometer with 1 GHz signal bandwidth, does notneed the optical surface of the collectors to be much more accurate than 3 cm and the pointing doesnot need to be controlled to better than a few arc minutes for 10 m telescopes. Interestingly, these5 = ∆ A ∆ t C A BC S t a r θθ θ ∆ B t Figure 1: Considering two harmonic emitting points in the source, with slightly different frequencies,telescopes A and B being close together receive beats almost at the same time. This corresponds toa high degree of correlation. Telescopes A and C being further apart do not receive beats in phaseand the degree of correlation is then lower [21].primary design requirements are well matched by Imaging Atmospheric Cherenkov telescopes usedfor gamma-ray astronomy above 100 GeV. SII also appears to be the only viable tool for highresolution studies through the atmosphere in blue wavelengths. Observations at short wavelengthswill provide a higher contrast for measurement of spatial differences in the envelope of hot stars.These wavelengths can potentially extend imaging capabilities of bright stars to larger distancesthan Michelson interferometry.
The intensity interferometry technique was initially developed for radio astronomy [17]. HanburyBrown and Twiss subsequently demonstrated that the technique could also be employed at visiblewavelengths [15, 16]. Based upon the success of these initial measurements, the Narrabri StellarIntensity Interferometer (NSII) was designed and constructed [14]. NSII was located near Narrabri,Australia, and operated from 1965 to 1976 (Figure 2). The NSII consisted of two telescopes, each6.5 m in diameter, with an 11 m focal length. The two telescopes were carried on trucks runningon a circular railway track 188 m in diameter. This allowed the interferometer to operate with abaseline of 10 m to 188 m. The combination of mirror area, electronic bandwidth, and electronicnoise restricted observations to stars brighter than B = +2.5. Still, the NSII was the first instrumentto allow the successful measurement of angular diameters of main sequence stars. A total of 32angular diameters were measured with the NSII, some as small as 0.4 mas [13].Imaging Air Cherenkov Telescopes (IACT) are used for gamma-ray astronomy at Very HighEnergies (VHE, energies greater than 100 GeV). The IACT technique relies on the fact that VHEparticles and gamma-rays initiate extensive air showers of high energy secondary particles in theatmosphere. Charged shower particles with sufficient kinetic energy will radiate optical Cherenkovlight into the atmosphere. This Cherenkov light at ground level is strongly peaked at blue wave-lengths, has a duration of only a few nanoseconds, and is also very faint ( ∼
10 photons / m at100 GeV). Large ( > m diameter) light collectors with excellent U/V band reflectivity equippedwith fast electronics are employed to detect this Cherenkov light. IACTs are typically used inwidely-spaced arrays of 2 to 4 telescopes with typically 100 m inter-telescope distances in order torecord stereoscopic views of each shower. This telescope separation is chosen to match the extentof the Cherenkov light pool at ground level [38], and is dependent on the altitude of the IACT6 (cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:0)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)(cid:1) P h o t o d e t e c t o r C o lli m a t o rI n t e rf e r o m e t r i c f ilt e r C o n c e n t r a t o r From telescope dishf/D = 1 D i a p h r a g m M i rr o r Gamma ray camera
Figure 2: The two 6.5 m Narrabri Stellar Intensity Interferometer telescopes (top left) have manycommon points with Imaging Air Cherenkov Telescopes such as the two 17 m MAGIC telescopes(center top), the four 12 m H.E.S.S. telescopes (right) or the four 12 m VERITAS telescopes (bot-tom). On the right is a schematic of a possible implementation of SII on a Cherenkov telescope[22]. The light from the mirrors is reflected side way toward a collimator and analyzer before be-ing focused on the photo-detectors. Such a simple system can be mounted on the shutter of theatmospheric Cherenkov camera so it does not block the view for gamma-ray observations.observatory, Figure 2 presents examples of current IACT telescopes.Because the design requirements of IACT arrays and SII arrays are generally similar, thesefacilities can potentially be used for both types of astronomical observations [22, 20]. Historically, thetwo SII telescopes of the Narrabri interferometer were subsequently instrumented as air Cherenkovtelescopes in a search for astronomical sources of very high energy (E >
300 GeV) gamma-rays[10, 11].An existing IACT telescope can be converted into a SII receiver by adding an optical collimator,an interferometric filter, a photo-detector and front end electronics on the focal plane of the IACTtelescope. The SII instrumentation could be mounted in front Cherenkov cameras as illustratedon the right side of Figure 2. It is desirable to have the SII instrument package integrated intothe camera shutter so that it does not interfere with normal gamma-ray observations. The HESStelescopes have already prototyped such a mounting system for performing photo-metric measure-ments of stars with a photomultiplier tube [4]. The SII instrument will send the SII signal fromeach telescope to a central facility to be combined and correlated over various baselines, therebyfilling out the u-v interferometric plane. The length and orientation of the effective baseline betweentwo IACT telescopes depends on the position of the star in the sky. At the time of the final dataanalysis, the interferometric (u,v) plane is binned, taking into account the time-varying baselinesfrom each telescope pair.Although IACT telescopes can be used for SII, IACT arrays are optimized for high energygamma-ray astronomy. Optimizations for highest gamma-ray sensitivity with lowest cost will con-strain the SII capabilities of these instruments. For example, the signal to noise ratio in an SIIinstrument depends upon the square root of the correlator bandwidth. IACT telescope designs typ-ically employ a large ( > ◦ diameter) field of view in order to capture the entire length of the opticalCherenkov signal as well as to efficiently map the night sky for new sources of VHE gamma-rays.The requirement to minimize optical aberration over the wide field of view has led several groups toadopt the Davies-Cotton [3] telescope design. However, this design does not preserve iso-chronicity(simultaneous arrival at the focal plane of all photons collected by single flash of light across thefull mirror surface). For 12 m telescopes such as in HESS and VERITAS [36], the telescope opticsspreads an optical light impulse out to a duration of several nanoseconds at the focal plane. This7
00 m m Figure 3: The proposed lay-out for the future CTA (left) includes 85 100 m dishes (red dots) andfour 600 m dishes (blue dots). It would offer 3,916 baselines [20] ranging from 35 m to 1,414 m(center) and allowing imaging at 400 nm over an aliasing free 1.6 mas field of view with a betterthan 0.1 mas angular resolution (right).effectively convolutes any optical signal with an effective filter bandwidth of ∼
100 MHz. Withparabolic telescopes such as in MAGIC (17 m) or H.E.S.S. II (29 m), the effective signal bandwidthfilter is on the order of 1 GHz.The optical quality of the IACT telescope will also constrain the limiting visual magnitude forSII applications. The optical point spread function of typical Davies-Cotton IACT design is 0 . ◦ .This point spread function is well matched to the typical angular width of an air Cherenkov lightimage (0 . ◦ ). However, because of the corresponding integration of night sky background light,these telescopes could not be used for SII observations of stars any fainter than visual magnitude ∼ . ∼
100 MHz system requires time delay correction accuracyof ∼ . ∼
10 nights each month during near-full moonlight are unusedfor gamma-ray astronomy, and could be made available for SII observations. This corresponds to ∼
800 hours of SII observation per year. Assured telescope access of this magnitude is more thansufficient for development of a vigorous SII science program. At the same time, the SII programwill have minimal impact on existing VHE gamma-ray science programs. If the SII science programis successful using the freely available 800 hours/year on the next generation IACT arrays, it wouldbe advantageous to consider building a future dedicated SII facility with a square kilometer arrayon varying, non-periodic baselines. The first-light time-frame for a dedicated facility would be inthe 2020-2030 decade, or later. 8igure 4: One the left, four examples of reconstructed images from a simulated SII observation usingan IACT telescope array. The simulation uses a wavelength of ∼
400 nm with one hundred CTA-liketelescopes and a inter-telescope separation of ∼
100 m. The pristine image is shown at the top leftin each example. The images were produced [28] using an algorithm based on the Cauchy-Riemannequations [19]. The analysis does not yet include a realistic noise component, which is still activelybeing investigated. One the right, for a range of degrees of coherence | γ | , the visual magnitudefor which a five standard deviation measurement is possible with a single baseline is indicated asa function of the observing time. We assumed A = 100 m light collectors with quantum efficiency α = 30% and a signal bandwidth of 100 MHz. Baseline multiplicities of 10, 30, 100 and 300 improvethe sensitivity by 1.25, 1.85, 2.51 and 3.11 magnitudes respectively. This is for model independentimaging only. Model fitting would allow the study of objects with fainter visual magnitudes. Next generation IACT arrays such as CTA [2] and AGIS [1] are planned for construction andoperation during the decade 2010-2020. These square kilometer arrays will use up to 100 telescopes(Figure 3) providing up to 5000 base-lines ranging from ∼
50 m to ∼ . / N) is(
S/N ) RMS = A · α · n ( λ ) | γ ( d ) | q ∆ f · T / α is the quantum efficiency, | γ (d) | the square9f the magnitude of the degree of coherence for a given baseline d. The right side of Figure 4 showsthe expected S / N ratio as a function of the exposure time for conservative telescope parameters(A = 100 m , α = 30% and ∆f = 100 MHz). With five hours of observation, a single baselinecan provide measurements of | γ | = 0 . | γ | = 0 .
03 for stars of visual magnitude 4.8 and 2.4respectively, with a 5 σ statistical significance.The statistical sensitivity of these observations can be further improved by 1 to 3 magnitudes(depending of number of baselines used and specific array design) by exploiting the redundancy ofthe baseline measurements in the full array. This could be further improved if additional opticalchannels are available at each telescope. Finally, when the information sought does not require amodel independent reconstruction of the image (as in the case of a binary, a fast rotating oblatestar, a star with an obstructing disk, etc), all the baselines can be simultaneously exploited to pushSII sensitivity to objects approaching m v = 9.For typical Cherenkov telescope quality (PSF = 0 . ◦ , area = 100 m ), the ultimate sensitivity (m v = 9) is dominated by the integrated night sky background in the PSF, and not by the SII signalstatistics. The observation of objects fainter than m v = 9 requires both higher telescope opticalquality as well as extended observation time. For each additional visual magnitude, the requiredobservation times increases by a factor >
6. This of does not preclude the utilization of largertelescopes (Area >
100 m ) as long as the source remains unresolved across the diameter of themirror. In particular, a five hour observation by a single pair of CTA 600 m telescopes will provide | γ | = 0 . | γ | = 0 .
03 measurements for visual magnitude 6.7 and 4.3 stars, respectively.
The currently envisioned SII implementation does not depend on the development of any criticaltechnological capability. There are technology developments, however, that can benefit the ca-pabilities of future large scale Intensity Interferometer arrays. These technological developmentsmay extend SII to higher sensitivities, thereby extending the imaging capabilities to dimmer visualmagnitudes. In this section, we describe the potential impact of two technological drivers.
A major improvement in the ultimate sensitivity of a large SII array can be accomplished by con-tinuously recording each telescope’s stream of intensity fluctuations with minimal signal processing,and calculating the inter-telescope correlation functions post-observation. This capability can al-low studying the degree of correlation as a function of time-lag between the two photon streams.Ofir and Ribak [30] have recently proposed that the time-lag dependence of the correlation carriesdetailed information regarding the source emission process. The availability of the full data streamfrom each telescope can also allow analysis techniques based on higher order intensity correlations(i.e. 2, 3 and more points). These higher order correlations are employed for both phase recon-struction analysis and also for absolute source distance measurements based on light wave frontcurvature [32, 30].The ability to handle such large data rates has only recently become technically feasible. Acontinuous 200
M Hz digitization of the photo-detector signal with 4 bit accuracy generates 360 Gb per hour of operation of a single CTA telescope, and about 280 T b per observation night for thefull array. Using existing technology such as off the shelf 1.5 Tb SATA disks (2009 price : $150each) in a distributed, parallel RAID array, it is feasible to archive the data in real time for next-day processing. The next day analysis pipeline would employ a farm of distributed processors togenerate the intensity correlations and invert the Fourier plane to generate the source image for10ach observation. The data storage requirements are minimal once the SII data has been correlatedand inverted (10-100 Gbyte/night maximum), and so after analysis, the RAID data storage farmwould be overwritten by the next night’s data. As hard-disk technology continues to advance, itmay eventually be possible to archive the entire raw data set each night for longer term study andre-analysis.Technological challenges associated with a 1-day turnaround pipeline analysis are of similarmagnitude to other state-of-the-art telescope systems such as LOFAR, ALMA, and SKA. The mostsignificant technological challenge to a large SII array is probably in establishment of sufficientbandwidth across the widely distributed array that will allow efficient transport all data to localizeddata storage nodes for collection and correlation. The data collection and processing scheme must bedesigned to provide sufficient redundancy and robust operation so that it may continue to performits analysis task even when one or more telescope systems in the array have failed.
As described previously, the
S/N ratio for SII is proportional to the photo-detector quantum effi-ciency α and also to the square root of the combined photo-detector/optical bandwidth ∆ f . For afixed stellar magnitude, improvements in the S/N ratio (and therefore visual magnitude sensitiv-ity) can therefore be achieved through technological improvements to the photo-detector quantumefficiency and speed (Figure 5).A standard bialkali photomultiplier tube will provide quantum efficiency α = 25% at opticalwavelength λ = 350 nm and signal bandwidth ∆ f = 200 M Hz . Super-bialkali photomultipliertubes have recently become available with α = 35%, and Ultra-bialkali with α = 43%. Conse-quently, one may improve the S/N ratio by a factor of two for a minimal cost per telescope. Thesensitivity improvement is equivalent to doubling the mirror area with a standard bialkali photo-multiplier tube. The application of ultra-bialkali photomultipliers to SII is fairly straightforwardand relatively inexpensive. This technology has yet to be deployed on a current SII instrument;The SII observation environment poses special challenges due to the strong optical intensity at thefocal plane. Issues regarding temperature stability, saturation, and photomultiplier tube lifetimehave yet to be explored. This technological development is underway and should proceed rapidlyover the next few years.The highest quantum efficiency photo-detectors commercially available are based upon semi-conductor technology; these devices (e.g. avalanche photodiodes) may approach α = 95% at λ = 330 nm. Large area silicon devices suffer from substantial inherent electronic noise relatedto the silicon bulk resistivity and the depth of the depletion layer at the PN junction. In order toachieve single photon sensitivity, recent technology has focused on the development of large arraysof very small area Single Photon Avalanche Photodiodes (SPADs). By minimizing the SPAD area(and hence the capacitance), each SPAD achieves sufficiently low inherent electrical noise to allowdetection of single photons with α > S/N ofa factor of 3-6 over conventional photomultiplier tubes, and 2-4 over ultra-bialkali photomultipliertechnology.Present technology SPADs are relatively inexpensive, and are stable with respect to intense lightexposure, but suffer from comparatively large (80 nsec) dead-time after observation of a light pulse.At the present time, a first implementation of a high speed, high quantum efficiency camera usingSPAD technology has been constructed and tested. This device, called AquEYE [27], is a firstgeneration implementation the ESO QuantEYE concept[26]. AquEYE employs four independent11igure 5: Sensitivity contour lines are separated by one visual magnitude in the signal bandwidthversus quantum efficiency plane characterizing photo-detectors. SPADs could give a 4 magnitudeimprovement compared to regular bialkali photomultipliers. However, the effective signal bandwidthif limited to 100 MHz for 10 m Davis-Cotton f/1 Cherenkov telescopes. Larger Cherenkov telescopes(MAGIC, H.E.S.S. II) are parabolic and provide a signal bandwidth close to 1 GHz.50 µ m diameter SPAD pixels, with pixel timing better than 50 psec and α >
50% for λ between500 nm and 600 nm. AquEYE was first successfully tested in June 2007 on the 182 cm Asagio CimaEkar Telescope. Future technological development of SPAD type photo-detector technology willneed to focus on reducing detector dead-time and overall system cost. The main organizational body overseeing the development of Intensity Interferometry as a modernastronomic technique is the SII Working Group within IAU commission 54 on optical and infraredinterferometry. The SII Working Group, commissioned in 2008, held its initial meeting at theAugust 2008 SPIE meeting in Marseilles, France. At this meeting, it was decided to organize an SIIworkshop in January 2009 at the University of Utah in order to review the scientific and technicalstatus of international SII development activities, and to coordinate research efforts. Members ofthe SII working group are independently pursuing science topics and technical developments relatedto SII. Some of these activities include: • Correlators, optical fibers, and photo-detectors are being developed and tested by independentresearch groups in the US and Europe. • The University of Utah is deploying a pair of 3 m telescopes on a 20 m East-West baseline in order to provide a realistic SII test environment for the SII working group. • SII Working Group Members of the VERITAS collaboration, in collaboration with Dravins,have performed inter-telescope correlation feasibility studies in October 2008 at the VERITASObservatory. Correlations between fast intensity fluctuations between two VERITAS 12 m ∼ lebohec/SIIWGWS/ • Close contacts are maintained with the CTA and AGIS collaboration regarding SII sciencetopics and integration of SII specific electronics and secondary optics with the IACT gamma-ray cameras. • A working group focused specifically on SII science and SII camera integration, led by Dravins,has been established with the CTA collaboration. The CTA SII group has recently submitteda CTA Phase A design study for implementation of SII capability onto the CTA Observatory. • Theoretical studies of optimal image and phase reconstruction techniques are underway, withsignificant efforts in Israel, Utah, and Europe.Ongoing progress in all of these various directions necessitates frequent meetings to discuss results,ideas, and to plan new initiatives. The SII Working Group also serves as the main organizationfor development of the research collaboration that will develop the SII instrumentation and scienceresearch programs for the CTA and/or AGIS Observatories. We intend to coordinate these variousefforts through yearly meetings of the SII Working Group.
The SII working Group envisions the development of the SII in four distinct Phases1. Initial Prototype Phase (2008-2010)2. IACT Array ‘Piggyback’ Phase (2011-2013)3. CTA/AGIS Science Program Phase (2013-2020)4. Dedicated SII facility phase (2020-2030)During the Initial Prototype Phase (2008-2010), the SII working group is working on parallel andcomplementary developments of experimental and computational tools and facilities that will allowinitial observations using a modern SII system. Part of this work will involves using the GrantsvilleSII 3 m telescopes to test out system components and reproduce measurements of stellar diametersand binaries. This phase will generate initial experience and validation of the SII technique whichwill prepare the SII Working Group to deploy a multiple telescope ( >
2) SII system.Experience and technology developed during Phase 1 will be refined and standardized duringPhase 2 which will likely include the instalation and testing of a SII system at a Current IACTarray. The SII working group will work with the VERITAS and HESS collaborations to develop afully instrumented SII system than can be deployed on either or both IACT arrays. The SII groupwill submit science and development proposals to these observatories to deploy and operate a fullfour telescope SII system. We anticipate performing 2 years of science observations with the SIIsystem during moonlight conditions, accumulating up to 1600 hours of source observations duringthis time. During Phase 2, an emphasis will also be placed on developing a reliable SII systemthat can be fabricated and deployed on the larger CTA/AGIS arrays, which should then be underconstruction.In 2013, we will enter Phase 3 which includes operating a 50-100 telescope SII system. Whenthe first CTA/AGIS telescopes are deployed, a small subset will be equipped with SII hardware forfurther testing (2013-2015) while the full scale implementation will be under preparation. By 2016,13he entire CTA array will have been equipped with SII hardware and the observation program willbe carried until 2020. The large number of redundant baselines sampled will substantially extendthe capabilities of the SII technique. It should be possible to establish a strong science programwith excellent image reconstruction and angular resolution in the sub micro-arcsecond range.Phase 4 would include the construction and operation of a dedicated SII km area array of highsignal bandwidth ( ∼ GHz) telescopes. This would allow a substantial increase in the number ofavailable observation hours especially in dark sky (no moonlight) conditions. This should result inincreased sensitivity for observing more distant (fainter) objects. This stage would take advantageof experience gained with CTA/AGIS in the operation of large arrays of large light collectors. Thisproject is beyond the current decanal survey.
As the proposed SII observations can be carried out using existing technology, the cost of many ofthe key items for the SII project can already be reliably estimated. However, the basic design ofthe AGIS and CTA telescopes is still under discussion, including the number of telescopes, separa-tions, diameters, etc. Consequently, the cost of the presently envisioned full SII implementation onCTA/AGIS inherits cost estimate uncertainties from uncertainties in the design of these observa-tories. We can make a representative estimate of the SII implementation based upon a strawmanCTA Observatory. The strawman CTA Observatory employs 97 telescopes which will generate 4656baselines. Each CTA telescope will be equipped with secondary SII optics ( ∼ $5 , ∼ $3 , M Hz bandwidth. Thiscan be achieved by optical fibers running to each telescope ($2 , , , , , References
L47[6] de Wit, W.J., Lamers, H.J.G.L.M., Marquette, J.B., et al. 2006 A&A
L1[8] Fontana, P.R., 1983, J. Appl. Phys., 54, 2, 473[9] Gamo, H., 1963, J. Appl. Phys., 34, 4, 875[10] Grindlay, J. E., et al., 1975, Astrophysical Journal, 201, 82[11] Grindlay, J. E., et al., 1975, Astrophysical Journal, 197, L9[12] Groenewegen, M.A.T., 1999, A&AS , 697[20] LeBohec, S., et al., 2008a, in The Universe at Sub-second timescales, Ed. D.Phelan, O.Ryanand A.Shearer, AIP Conf. Proc. 984, 205[21] LeBohec, S. et al., 2008b, Proc. SPIE Vol. 7013, 70132B[22] LeBohec S. and Holder, J., 2006, ApJ 649, 399[23] Labeyrie,A., Lipson,S.G., Nisenson,P., 2006, An Introduction to Optical Stellar Interferome-try, Cambridge Univ. Press, Cambridge[24] Marathay, A.S., Hu, Y.Z. and Shao, L., 1994, Opt.Eng., 33, 10, 32651525] Merand et al, 2005A&A 438L, 5[26] Naletto, G. et al., 2006 QuantEYE: a quantum optics instrument for extremely large tele-scopes, Proceedings of the SPIE , Ground-Based and Airborne Instrumentation for As-tronomy, Ian S. Mclean and Masanori Iye , eds. , Proceedings of the SPIE , 62691W.[27] Naletto, G. , C. & al., 2007, Very fast photon counting photometer for astronomical appli-cations: from QuantEYE to AquEYE, Photon Counting Applications, Quantum Optics, andQuantum Cryptography. Edited by Prochazka, I et al., Proceedings of the SPIE, Volume , 65830B[28] Nunez P, Le Bohec S, et al. 2009 in prep.[29] P. Lawson, 2008, Optical Long Baseline Interferometry News, http://olbin.jpl.nasa.gov/[30] Ofir A. and Ribak, E., 2006, M.N.R.A.S. , 1646-51.[31] Pare, E. et al., 2002, NIM A 490, 71[32] Jain P. and Ralston J. P., 2008, A&A 484, 887-895.[33] Sasselov, D. and Karovska, M., 1994 ApJ
American Astronomical SocietyMeeting Abstracts42