© ESO - September 2004 17

TThe ESO STC recommended the construction ofSINFONI (“Spectrograph for INtegral FieldObservation in the Near-Infrared”) in October 1997.The instrument was conceived as the combinationof an Adaptive Optics facility (the AO-Module),

developed by ESO, and SPIFFI (SPectrometer for Infrared FaintField Imaging), a Near Infrared Integral Field Spectrograph devel-oped by the MPE. It merges in one instrument the experience gainedby ESO in Adaptive Optics (AO) with the development of theMACAO product line and the experience of MPE in integral fieldspectrometry with 3D, the world’s first infrared integral field spec-trometer.

A two year detailed design phase was initiated in 2000. At thefinal design review (2001), the board recommended the upgrade ofthe instrument to a higher spectral resolution (nearly 4000) with a2048 by 2048 pixel infrared detector. NOVA joined the consortium atthat time to undertake the development of the so-called ‘2K Camera’scaling the spectral images to the size of the new detector.

FFUNCTIONALITIESUNCTIONALITIES

The AO correction is based on the analysis of the wavefront of a ref-erence star picked up close to the science Field of View (FOV). Thesharing of the Cassegrain FOV between the two sub-systems is real-ized by an infrared dichroic mounted at the entrance of the SPIFFIcryostat. The visible flux (450 to 1000 nm) is reflected towards thewavefront sensor and the infrared (1.05 to 2.45 µm) is transmitted tothe science focal plane.

The AO reference star can be acquired up to one arcmin off-axis.

The optimum performance is achieved with stars brighter thanR = 12 mag picked up within ~20 arcsec but fair correction can beobtained in good atmospheric conditions over the full 1u2 arcmin2

FOV with stars up to R = 17 mag. The AO-Module also supports see-ing-limited operations when no guide star is available.

FFIRSTIRST LLIGHTIGHT OFOF SINFONISINFONI AATT THETHE VLVLTTSINFONI IS AN ADAPTIVE OPTICS ASSISTED CRYOGENIC NEAR INFRARED SPECTROGRAPH DEVELOPED BY

ESO AND MPE IN COLLABORATION WITH NOVA. IT WAS SUCCESSFULLY COMMISSIONED BETWEEN JUNE AND

AUGUST AT THE CASSEGRAIN FOCUS OF VLT UT4 (YEPUN). THE INSTRUMENT WILL BE OFFERED TO THE COM-MUNITY FROM APRIL 2005 ONWARDS (PERIOD 75) AND WILL PROVIDE A UNIQUE FACILITY IN THE FIELD OF HIGH

SPATIAL AND SPECTRAL RESOLUTION STUDIES OF COMPACT OBJECTS (STAR-FORMING REGIONS, NUCLEI OF

GALAXIES, COSMOLOGICALLY DISTANT GALAXIES, GALACTIC CENTRE ETC.).

HHENRIENRI BBONNETONNET 11, R, ROBEROBERTOTO AABUTERBUTER 22, A, ANDREWNDREW BBAKERAKER 22, W, WALALTERTER BBORNEMANNORNEMANN 22, A, ANTHONYNTHONY

BBROWNROWN 88, R, ROBEROBERTOTO CCASTILLOASTILLO 11, R, RALFALF CCONZELMANNONZELMANN 11, R, ROMUALDOMUALD DDAMSTERAMSTER 11, R, RICHARDICHARD DDAAVIESVIES 22,,BBERNARDERNARD DDELABREELABRE 11, R, ROBOB DDONALDSONONALDSON 11, C, CHRISTOPHEHRISTOPHE DDUMASUMAS 11, F, FRANKRANK EEISENHAUERISENHAUER 22, E, EDDIEDDIE

EELSWIJKLSWIJK 99, E, ENRICONRICO FFEDRIGOEDRIGO 11, G, GERERTT FFINGERINGER 11, H, HANSANS GGEMPERLEINEMPERLEIN 22, R, REINHARDEINHARD GGENZELENZEL 2,32,3,,AANDREANDREA GGILBERILBERTT 22, G, GORDONORDON GGILLETILLET 11, A, ARMINRMIN GGOLDBRUNNEROLDBRUNNER 22, M, MAATTHEWTTHEW HHORROBINORROBIN 22, R, RIKIK TERTER

HHORSTORST 99, S, STEFTEFANAN HHUBERUBER 22, N, NORBERORBERTT HHUBINUBIN 11, C, CHRISTOFHRISTOF IISERLOHESERLOHE 2,42,4, A, ANDREASNDREAS KKAUFERAUFER 11,,MMARKUSARKUS KKISSLERISSLER-P-PAATIGTIG 11, J, JANAN KKRAGTRAGT 99, G, GABBYABBY KKROESROES 99, M, MAATTHEWTTHEW LLEHNEREHNERTT 22, W, WERNERERNER LLIEBIEB 22,,

JJOCHENOCHEN LLISKEISKE 11, J, JEANEAN-L-LOUISOUIS LLIZONIZON 11, D, DIETERIETER LLUTZUTZ 22, A, ANDREANDREA MMODIGLIANIODIGLIANI 11, G, GUYUY MMONNETONNET 11,,NNICOLEICOLE NNESVESVADBAADBA 22, J, JONAONA PPAATIGTIG 11, J, JOHANOHAN PPRAGTRAGT 99, J, JUHAUHA RREUNANENEUNANEN 11, C, CLAUDIALAUDIA RRÖHRLEÖHRLE 22, S, SILILVIOVIO

RROSSIOSSI 11, R, RICCARDOICCARDO SSCHMUTZERCHMUTZER 11, T, TONON SSCHOENMAKERCHOENMAKER 99, J, JÜRGENÜRGEN SSCHREIBERCHREIBER 2,52,5, S, STEFTEFANAN

SSTRÖBELETRÖBELE 11, T, THOMASHOMAS SSZEIFERZEIFERTT 11, L, L INDAINDA TTACCONIACCONI 22, M, MAATTHIASTTHIAS TTECZAECZA 2,62,6, N, NIRANJANIRANJAN TTHAHATTETTE 2,62,6,,SSEBASTIENEBASTIEN TTORDOORDO 11, P, PAULAUL VVANAN DERDER WWERFERF 88, H, HARALDARALD WWEISZEISZ 77

1EUROPEAN SOUTHERN OBSERVATORY; 2MAX-PLANCK-INSTITUT FÜR EXTRATERRESTRISCHE PHYSIK, GARCHING, GERMANY;3DEPARTMENT OF PHYSICS, UC BERKELEY, USA; 4UNIVERSITÄT KÖLN, GERMANY; 5MAX-PLANCK-INSTITUT FÜR ASTRONOMIE,

HEIDELBERG, GERMANY; 6OXFORD UNIVERSITY, UK; 7INGENIEURBÜRO WEISZ, MÜNCHEN, GERMANY; 8LEIDEN OBSERVATORY,NOVA, THE NETHERLANDS; 9NETHERLANDS FOUNDATION FOR RESEARCH IN ASTRONOMY

Figure 1: Principle of SPIFFI.

SPIFFI obtains spectra for each spatialelement of its two-dimensional FOV. Itsfunctional principle is shown in Fig. 1. Aftersome processing of the raw data the outcomeis a data cube, with 2 spatial (64 by 32 pix-els) and one spectral (2048 pixels) dimen-sions. Spatially, three magnifications areoffered. The high-resolution mode, with 25masu 12.5 (milli-arcsec) per pixel, providesa Nyquist sampling of the diffraction-limitedPoint Spread Function (PSF) at 2 µm with anarrow FOV (0.8u0.8 arcsec2). This modewill offer spatially resolved spectral data oncosmologically distant galaxies, but will belimited by readout noise in many applica-tions even in the case of long integrations(30 minutes). The intermediate pixel scale(100 u 50 mas/pixel) offers the best com-promise between FOV, sensitivity and spa-tial resolution for most applications. Thecoarse mode (250 u 125 mas/pixel) is suit-able for seeing-limited operations and also,thanks to its “wide” 8u8 arcsec2 FOV, isvery useful to help identifying the target dur-ing the acquisition.

Four spectroscopic modes are support-ed: the J-, H- and K-band gratings provide aspectral resolution of approx. 2000–4000. Alower resolution (~2000) is achieved withthe combined H and K grating.

IINSTNSTALLAALLATIONTION ANDAND

COMMISSIONINGCOMMISSIONING SEQUENCESEQUENCE

The integration of SPIFFI with the “1KCamera” was completed at the beginning of2002 in a configuration enabling a directinterface to the VLT for seeing-limitedobservations. SPIFFI was then operated as aguest instrument without AO for a period ofthree months, producing outstanding scien-tific data (see The Messenger 113, 17) aswell as valuable experience in preparationfor the AO-assisted operations. The instru-

ment was then shipped back to Europe andconditioned to interface to the AO-Module.The integration of the latter and its perform-ance demonstration in the laboratory werecompleted in December 2003. The two sub-systems were coupled for the first time in theESO Garching integration facility in January2004 and the Preliminary Acceptance inEurope was awarded in March. The 2KCamera and Detector assembly was thenintegrated in SPIFFI while the AO-Modulewas shipped to Chile for re-integration andstand-alone commissioning on the sky.

The first “AO” light was obtained onMay 31st. On this occasion, an Infrared TestCamera was mounted at the AO-Modulecorrected focus to provide direct imagingcapabilities (Fig. 2). The adaptive opticsloop was immediately closed on a bright star(mV ~ 11) and provided a corrected imageconcentrating into a diffraction-limited core56% of the energy in the K-band (λ = 2.2µm,Figure 3). This performance conformed toexpectations based on laboratory experi-

ments and the experience obtained with thetwo MACAO systems that had already beentested at the Coudé foci of the VLTs.

The re-integration of SPIFFI with theAO-Module took place in the assembly hallof Paranal in June and SINFONI first lightwas obtained on July 9 (Fig. 4). The firstcommissioning was devoted to the testing ofall the observing modes and the tuning of themany automatic control loops which willultimately make this complex instrumentuser-friendly. The success of the July runallowed for dividing the August commis-sioning run among commissioning activitiesfor fine-tuning operations, ScienceVerification and Guaranteed TimeObservations.

OOPENINGPENING TOTO THETHE COMMUNITYCOMMUNITY

The instrument is currently equipped with anengineering-grade detector that is suitablefor technical qualification but affected byflaws which would limit the science capabil-ities of SINFONI. The final science-gradedetector is currently undergoing characteri-zation tests in Garching and will be integrat-ed in the instrument in November.

Meanwhile, the first call for proposals tothe astronomical community will be issuedin September and the start of regular obser-vations is scheduled for Period 75 (startingApril 1st 2005).

Adaptive Optics correction requires thepresence of a reasonably bright referencestar in the vicinity of the science target. Thisconstraint limits AO operations to a smallfraction of the sky. This limitation will beovercome as soon as Yepun (UT4) isequipped in 2005 with a Laser Guide StarFacility (LGSF). The LGSF will launch a10W Sodium laser beam along the telescopeline of sight to create an artificial beacon inthe high atmospheric sodium layer(~100km), providing a reference for wave-front sensing. The upgrade of SINFONI tolaser guide star operations is foreseen for thesecond quarter of 2005.

The Messenger 11718

Figure 2: TheAO-Module atthe VLT UT4Cassegrainfocus for itsstand-alonecommission-ing. The AOoptical benchis the light-blue ring. Itinterfaces tothe VLTCassegrainflangethrough theAO-Housing(intermediatering) and theinterfaceflange (topdark ring).The yellowtest structuresimulates the mass of SPIFFI and provides the mechanical interface to the Infrared TestCamera (aluminum box in the yellow ring).

Figure 3: AO-Module firstlight. The magnitude of thestar is ~11 in V, the seeing0.65 arcsec. The measuredStrehl is 56% (λ = 2.16 µm).The displayed FOV is 0.5arcsec, with a sampling of15 mas/pixel.

IINSTRUMENTNSTRUMENT OOVERVERVIEWVIEWTTHEHE AADAPTIVEDAPTIVE OOPTICSPTICS MMODULEODULE

Optical layoutThe AO-Module was developed in the ESOheadquarter laboratory in Garching (Bonnetet al. 2003). It consists in an optical relayfrom the VLT Cassegrain focus to theSPIFFI entrance focal plane, which includesa deformable mirror conjugated to the tele-scope pupil. The curvatures of the 60 actua-tors are updated at 420 Hz to compensate forthe aberrations produced by the turbulentatmosphere. The optical layout is shown inFig. 5.

At the reflected focus, a pick-up lensmounted to a tri-axial linear stage collectsthe visible flux of the AO guide star, whichis then imaged on the membrane mirror. Thewavefront sensor box (mirrors M1 to M4)collimates the beam and re-images the pupil(conjugated to the deformable mirror) on thelenslet array. There, the flux is split among60 sub-apertures and forwarded to a secondarray of lenslets, which image the sub-aper-tures on the cores of 60 optical fibers. Thephotons are then guided to 60 AvalanchePhoto Diodes (APDs) located in a cabinetmounted to the instrument housing.

In closed loop, a speaker located at theexit of an acoustic cavity forces the mem-brane mirror to oscillate at 2.1kHz. As aresult, the pupil plane moves back and forthalong the optical axis allowing the lensletarray to intercept alternatively the intra- andextra-focal regions. An aberration in thewavefront would de-collimate the beam

locally, causing the flux density to varyalong the optical axis. As APDs are read outsynchronously with the oscillations, the rel-ative intensity difference between the intra-and extra-focal phases is proportional to thelocal wavefront curvature. This signal isused to generate the corrective command tothe deformable mirror. The correction isupdated at 420Hz, with a closed-loop band-width of 30 to 60Hz. The control gains areoptimized for the actual atmospheric condi-tions and the guide-star brightness.

Additional opto-mechanical elementsare already integrated in the AO-Module inview of the Laser Guide Star mode opera-tions. They are not shown in the optical lay-out of Fig. 5.

Performance demonstration on the skyThe Infrared Test Camera (ITC) mounted tothe AO-Module during the stand-alone com-missioning run of June was used to demon-strate the performance of the AO-Module.The ITC provides direct imaging over a fair-ly large FOV (15u15 arcsec2) with a sam-pling adapted to the diffraction-limited PSFin the near infrared (15 mas/pixel).

The typical AO-corrected PSF of a starfeatures a diffraction-limited core superim-posed on a larger halo generated by uncor-rected high-order aberrations. The quality ofthe correction is evaluated with the Strehlratio, computed as the fraction of the incom-ing energy which falls into the diffraction-limited core. In addition to the star bright-ness, the Strehl ratio is driven by the ampli-tude and the coherence time (τ0) of the see-ing: in poor seeing conditions, the PSF isblurred by the larger amplitude of the high-order aberrations, while the AO-Modulereactivity becomes insufficient for fast see-ing conditions.

The measured performance of the AO-

© ESO - September 2004 19Bonnet H. et al., First Light of SINFONI

Figure 4: SINFONIduring installation atUT4. The SPIFFI cryo-stat vessel is the greyaluminum part belowthe AO-Module (blue).To the left, the M1 cellof the VLT.

Figure 5: Optical layout of the AO-Module.

Figure 6: Strehlmeasurements

obtained during AO-Module commission-

ing. Red squaresshow the specified

Strehl performance.Black (resp. blue) dots

are values obtainedwith τ0 > 3 ms (resp.

τ0 < 3 ms )

Str

ehl r

atio

Module is plotted vs. the guide-star magni-tude in Fig. 6. Black dots are Strehl ratiosobtained in conditions better than the medi-an Paranal atmosphere, i.e. with a seeing< 0.65) in the visible and a τ0 > 3 ms, forwhich the AO-Module was specified (speci-fications are shown by the red squares).Strehl ratios obtained in degraded conditionsare plotted in blue, for which the reductionin performance is larger than expected fromthe experience of the other MACAO sys-tems. We believe that they were caused by aloss of bandwidth of the instrument inducedby a variation of the membrane mirror gain

between the laboratory and the telescopeenvironmental conditions.

An illustration of the image qualityachieved in good conditions with a brightreference star is given in Figure 7, whichdisplays a pair of stars with similar magni-tudes (~7.5) and a separation of 0.75). Thebrightest component was used as the refer-ence star, with an absorbing filter dimmingthe flux by 2.5 magnitudes to avoid saturat-ing the detectors of the wavefront analyzer.The Strehl ratio cannot be estimated due tothe saturated peak of the bright component,but the image quality can be qualitatively

appreciated from the number of visible Airyrings.

The ability to perform AO correctionson faint guide objects is essential forSINFONI in order to enable observations offaint extragalactic objects without nearbybright reference stars. This capability wasdemonstrated for good seeing conditionswith a star of magnitude 17.5 (see Figure 8).The measured Full Width at Half Maximum(FWHM) of the corrected PSF in K-bandwas 0.145) while the uncorrected FWHMwas 0.38).

Figure 9 shows the reconstructed 3.6-day orbit of Linus, a satellite of Kalliope.Kalliope is a 180km-diameter asteroid orbit-ing the sun from within the asteroid mainbelt, between the orbits of Mars and Jupiter,at nearly three times the Earth-Sun distance(three astronomical units). The binary sys-tem was observed in early June 2004 overfour consecutive nights using the SINFONIAO module. Due to the geometry of thebinary system at the time of the observa-tions, the orbital plane of the satellite wasseen at an angle of 60 degrees with respectto our line of sight. In fact, the satellite(~50km diameter) revolves around Kalliopeon a 2000km-diameter circular orbit, with adirection of motion similar to the rotationspin of Kalliope (prograde rotation). Thediscovery of this asteroid satellite, namedLinus after the son of Kalliope, the Greekmuse of heroic poetry, was reported inSeptember 2001 by a group of astronomersusing the Canadian-France-Hawaii tele-scope in Hawaii (IAUC 7703). Additionalobservations of Kalliope have been obtainedsince 2001, which led to the determinationof the main orbital characteristics of thisbinary system. Although Kalliope was previ-ously thought to consist of metal-rich mate-rial, the discovery of Linus allowed scien-tists to derive a low ~2 g/cm3 mean densityfor Kalliope. This number is inconsistentwith a metal-rich object and Kalliope is nowbelieved to be a rubble-pile stony asteroid.Its porous interior is due to a catastrophiccollision with another smaller asteroidwhich occurred early in its history and gavebirth to Linus. The VLT data obtained withthe AO-Module of SINFONI will help refineour knowledge of the orbital characteristicsand structural properties of this binary sys-tem.

Finally, the most impressive image sofar was obtained during the August commis-sioning with SPIFFI: the Einstein Cross(Figure 10), a quadruple image of a singleQSO amplified by the gravitational lensinginduced by the foreground galaxy visible atthe centre of the cross. This H-band recon-structed image was obtained with 20 min-utes of integration time in the intermediateplate scale. The AO-Module was guiding onthe brighter southern component, with avisual magnitude of 16. The total extension

The Messenger 11720

Figure 7: Observation of a binary system with a separation of 0.75). To the left, open loopimage acquired with a seeing of 0.6 arcsec. To the right, closed loop image obtained in thesame conditions. The brighter component (lower right) served as a reference for the wave-front sensing. The additional “companions” at the lower left of the PSF cores are ghostimages produced by a multiple reflection at the entrance window of the Infrared TestCamera.

Figure 8: Performance ofAO-Module on faint

guide star (visual magni-tude 17.5). To the right,

the seeing-limited K-band image (FWHM 0.38

arcsec). To the left, theAO-corrected image

(FWHM 0.145 arcsec).The ability to perform AOcorrections on very faint

guide objects is essentialfor SINFONI observations

of faint extragalacticobjects.

Figure 9: Image of asteroidKalliope and its satelliteLinus. The frames wereacquired over four consecu-tive nights, surveying acomplete orbital revolutionof the system. The bright-ness of the central body(Kalliope) has been scaleddown by a factor 15 toreduce the contrast withLinus. Changes in the sizeof the unresolved Linusimage are due to variableseeing conditions from nightto night.

of the system is 1.8 arcsec, making it a verychallenging target for the wavefront sensordue to its complex structure and low bright-ness. The seeing conditions were excellentduring this integration (~0.5 arcsec with τ0 ~4ms).

TTHEHE NEARNEAR-- INFRAREDINFRARED INTEGRALINTEGRAL FIELDFIELD

SPECTROMETERSPECTROMETER SPIFFISPIFFISPIFFI was developed at the MPE(Eisenhauer et al. 2003) in collaborationwith NOVA and ESO. It provides imagingspectroscopy of a contiguous, two-dimen-sional field of 64 u 32 spatial pixels in thewavelength range from 1.1 – 2.45 µm at aresolving power of 2000 to 4000. As a result,the instrument delivers a simultaneous,three-dimensional data-cube with two spa-tial dimensions and one spectral dimension.Originally equipped with a 10242 pixeldetector, SPIFFI was upgraded with a newcamera unit built by NOVA and a HAWAII20482 RG detector array from Rockwell inthe two months between the SINFONI sys-tem tests in Garching and the commission-ing in Paranal. The integral field unit ofSPIFFI is a so-called mirror slicer (Tecza etal. 2002). This image slicer consists of twostacks of 32 plane mirrors, cutting the fieldof view into 32 narrow stripes and rearrang-ing these slitlets to form a single long slit.Depending on the selected image scale, theFOV of the integral field unit ranges from8) u 8) for seeing-limited observations to0.8) u 0.8) for imaging spectroscopy at thediffraction limit of the VLT. The intermedi-ate magnification with a 3.2) u 3.2) field ofview provides a compromise betweenreduced spatial resolution and improved sen-sitivity, and is best-suited for AO-assistedobservations of objects with low surfacebrightness. The slit width in the three imagescales is 250, 100, and 25 mas, with equiva-lent pixel sizes of 125, 50, and 12.5 mas.SPIFFI is equipped with four directly ruledgratings optimized for covering the atmos-pheric J, H, K, and combined H+K bands ina single exposure. The spectral resolvingpower for the 250 mas slit width is about

2000 in J, 3000 in H, 4000 in K, and 2000 inH+K. When using the adaptive optics imagescale, the spectral resolving power increasesup to about 2400 in J, 5500 in H, 5900 in K,and 2700 in H+K, but spectral dithering isnecessary to recover the full resolution fromthe undersampled spectra. The entire instru-ment is cooled with liquid nitrogen to a tem-

perature of –195° C. Figure 11 shows aninside view of SPIFFI.

A quick-look image reconstructionallows the instantaneous display of thereconstructed image during acquisition andobserving. SPIFFI is equipped with its owndata-reduction software, which is presentlyimplemented in the VLT data-reduction

© ESO - September 2004 21Bonnet H. et al., First Light of SINFONI

Figure 10: Reconstructed H-band image of theEinstein cross obtained with SINFONI using the100mas/pix mode (image size 3)x3)) under very

good seeing conditions (0.45)). This image demon-strates the capability of the MACAO AO system to

sharpen the images of faint astronomical objects(here the guide star is the brighter component andhas V=16 mag). The Einstein cross was discoveredin 1985 and is one of the best-known examples of

gravitational lensing. A massive black hole at thecentre of a nearby galaxy, located along the line ofsight to a distant quasar, deforms the geometry of

space in its immediate surroundings. As a result,the light from the quasar reaches us through differ-ent paths as if the quasar was observed through a

kaleidoscope, creating multiple images of the sameobject. This special optics distorts and magnifies

the light of background quasars, providing a meansof probing the distribution of intervening matter in

the cosmos.

Figure 11: An inside view of SPIFFI: the cryostat cover and the reinforcing structure havebeen removed to provide a view of the opto-mechanical components of SPIFFI. The lightenters from the top. The pre-optics with a filter-wheel and interchangeable lenses pro-vides three different image scales. The image slicer rearranges the two-dimensional fieldinto a pseudo-long slit, which is perpendicular to the base plate. Three diamond-turnedmirrors collimate the light on the gratings. In total, four gratings are installed on the grat-ing drive. A multiple-lens system then focuses the spectra on a Rockwell HAWAII array.In the meantime the spectrometer has been upgraded with an 20482 pixel detector anda new spectrometer camera. The diameter of the instrument vessel is 1.3 m.

pipeline. This software package provides alltools for the calibration and reduction ofSPIFFI data, including wavelength calibra-tion and image reconstruction. The final dataformat is a three-dimensional data cube with64 u 60 spatial pixels, and up to ~4500spectral elements.

Table 1 lists the limiting magnitudes forthe observation of a point source in the vari-ous instrument configurations.

FFIRSTIRST SCIENCESCIENCEFROMFROM COMMISSIONINGCOMMISSIONING

In addition to performance characterizationand optimization, the commissioning runswere used to explore the capabilities of theinstrument via test observations on a selec-tion of exciting astronomical targets.

DDIFFRACTIONIFFRACTION--LIMITEDLIMITED INTEGRALINTEGRAL--FIELDFIELD SPECTROSCOPYSPECTROSCOPY OFOF THETHE

GGALACTICALACTIC CCENTREENTRE

Because of its proximity of only 8 kpc, thecentre of the Milky Way is a unique labora-tory for studying physical processes that arethought to occur generally in galactic nuclei.The central parsec of our Galaxy contains adense star cluster with a remarkable numberof luminous and young, massive stars, aswell as several components of neutral, ion-ized and extremely hot gas. For two decades,evidence has been mounting that theGalactic Centre harbors a concentration ofdark mass associated with the compact radiosource SgrA* (diameter about 10 light min-utes), located at the centre of that cluster.Measurements of stellar velocities and (par-tial) orbits with the ESO NTT (SHARP) andVLT (NACO) have established a compellingcase that this dark mass concentration is amassive black hole of about 3.5⋅106 M0

(Schödel et al. 2002). The Galactic Centrethus presently constitutes the best proof wehave for the existence of massive blackholes in galactic nuclei. High-resolutionobservations offer the unique opportunities

to stringently test theblack-hole paradigm andstudy stars and gas in theimmediate vicinity of ablack hole, at a level ofdetail that will never beaccessible in any othergalactic nucleus. AO-assisted integral-fieldspectroscopy will play aunique role in studyingthis important region.

The Galactic Centrewas observed duringSINFONI commission-ing on July 15, 2004, forabout 2 hours in ~0.6)seeing conditions. TheAO-Module was lockedon a 14.6 mag star locat-

ed about 20) to the north-east of SgrA* andthe centre of the star cluster. We observedwith the finest plate scale (0.0125) u 0.025)per pixel) and the H+K grating (resolvingpower ~2000). Figure 12 (left insets) showsthe K cube collapsed into a quasi-continuumimage whose FWHM resolution is 75 mas,i.e. very close to the diffraction limit of theVLT in the K band. The Strehl ratio wasabout 10% in K band but is wavelengthdependent.

The SINFONI data provide for the firsttime a complete census of the near-IR spec-tra of most K ≤ 16 stars within about 20light-days of the black hole. The right upperinsets of Figure1 show that a number ofthese “S-stars” exhibit HI Brγ (and HeI)absorption, characteristic of early type, O–Bmain sequence stars. In fact we find that atleast 2/3 of all stars with K ≤ 16 in the cen-tral arcsecond have Brγ in absorption. Thisfinding increases dramatically the ‘paradoxof youth’ (Ghez et al. 2003), as to how these

massive and presumably young stars havemanaged to reside in the central few tens oflight days around the black hole. A numberof ideas have been put forward, including insitu formation in an extremely dense circum-nuclear gas disc, in-spiraling of a massiveyoung star cluster, perhaps aided by interme-diate-mass black holes, scattering by stellarblack holes, and collisional build-up of mas-sive stars from lower-mass stars, but none isso far fully convincing. The orbital proper-ties of the S-stars, now better constrainedwith the radial velocities obtained bySINFONI, promise to give important addi-tional clues to the origin of the massive cen-tral star cluster.

Additional SINFONI observations onJuly 16 also provide new information on thespectral properties of massive blue super-giants further away from SgrA*. An AO-assisted observation of IRS13 E (lower rightinset in Fig. 12), for the first time permittedspatially resolved spectroscopy of all threemajor components of this compact concen-tration of stars, which has been proposed tobe the remnant core of an in-spiraling youngstar cluster. We find that all these three com-ponents have the characteristics of differenttypes of Wolf-Rayet stars.

During the observations of July15 wehad the good luck to catch a small flare ofinfrared radiation from SgrA* itself (lowerleft inset in Fig. 12). These infrared flares,seen for the first time in NACO observationslast year (Genzel et al. 2003 b), probablysample in-falling hot gas in the immediatevicinity of the event horizon. By revealingthe spectral properties of this flare theSINFONI data offer key new insights intothe emission processes that cause infraredflares.

These few hours of data taken on theGalactic Centre during commissioning pro-

The Messenger 11722

Slit width 250 mas 100 mas 25 mas

J Band 18.0 18.5 17.0

H Band 17.5 18.5 17.3

K Band 16.9 18.0 17.2

H+K Band 17.7 18.9 18.2

Table 1: Limiting magnitudes for the observations of a pointsource. The sensitivities are calculated for a signal/noise of 10per spectral channel and one hour integration on source (6 x 10minutes, plus additional sky observations). We assume that50%, 50%, and 25% of the flux is encircled within a diameter of0.65), 0.2), and 0.1) for seeing-limited observations with the 250mas slitlets, and AO assisted observations with the 100 mas and25 mas slitlets, respectively.

Figure 12:Diffraction-limit-ed integral-fieldspectroscopy of

the GalacticCentre with

SINFONI. Thebackground

image shows acomposite

H,Ks,L’-band,three-colour

image taken withthe AO imager

NACO, delineat-ing the distribu-

tion of the differ-ent types of

stars and the hotinterstellar dust

in the centralparsec.

vide a small glimpse of the fantastic infor-mation that diffraction-limited spectroscopywill bring for Galactic Centre (and star clus-ter) research in the next few years. More willundoubtedly follow next year when theinstrument becomes available for scienceexploitation.

A A SHARPSHARP LOOKLOOK AATT

ACTIVEACTIVE GALACTICGALACTIC NUCLEINUCLEI

The role of star formation in active galacticnuclei (AGN) has long been a contentiousissue. Increasing evidence that starbursts dooccur in the vicinity of at least some AGNhas led to a change in thinking from the“starburst or black hole” debate of the last10–20 years to a new framework in whichthe fundamental questions concern thephysics linking the two phenomena. This ismore than just an issue of driving gas in tosmall scales because AGN and star forma-tion activity impact each other via feedbackmechanisms. But the extent of this interac-tion can only be ascertained by addressinghow common, how extended, how recent,and how energetically significant starburstsaround AGN are. Commissioning observa-tions of the Circinus galaxy and NGC 7469

demonstrate the potential of SINFONI toprobe the environment around AGN onsmall scales, and to reveal in detail the dis-tribution and dynamics of both the stellarand gaseous components.

The Circinus galaxy harbours a Seyfert2 nucleus and, at a distance of only 4 Mpc,is one of the nearest such objects. Althoughthe AGN itself is highly obscured, the com-pact K-band core also reveals the AGN indi-rectly, since it has a non-stellar origin: thedistribution of the stellar continuum isrevealed by the CO 2–0 absorption band-head at 2.29 µm to have a much broader dis-tribution. This more extended morphology isalso reflected in the H2 1–0 S(1) 2.12µmline, as reported by Davies et al. (1998), andthe Brγ 2.17 µm line. Because these lines arenarrow and show ordered rotation, it is like-ly that they arise in a disc around the AGN.The position angle and speed of the rotationare consistent with those at larger radii, indi-cating that down to scales of a few parsecsthere is no warp in the galactic plane.Circinus is known to exhibit a number ofhigh-excitation coronal lines. Of these, the[CaVIII] line at 2.32 µm (the ionisationpotential of which is 127 eV) is particularly

strong. An initial analysis of the line proper-ties reveals an increase in the velocity anddecrease in its dispersion from the nucleus toa region offset to the northwest. Our inter-pretation is that there are two components tothis line: a broad component centred on thenucleus at the systemic velocity and proba-bly associated with the Broad Line Region;and a narrow component, which is redshift-ed by an additional 140 km/s, offset in thedirection of the prominent ionisation cone,and likely associated with it.

NGC 7469 is a prototypical Seyfert 1galaxy at a distance 66 Mpc that is wellknown for its circumnuclear ring, as well asa significant contribution from star forma-tion to the nuclear luminosity (e.g. Genzel etal. 1995). Recent adaptive optics long-slitspectra combined with a millimeter interfer-ometric map of the cold gas (Davies et al.2004) led to a dynamical model in which thegas forms two rings, at 2.5) (800 pc) and 0.2)(65 pc). In this interpretation, the nuclearstar forming region, which contributes atleast half of the mass within 30 pc of thenucleus, was confined within the inner ring.These results are confirmed by new datafrom SINFONI which reveal the full 2-

© ESO - September 2004 23Bonnet H. et al., First Light of SINFONI

Figure 13: The Circinus galaxy - the near-est galaxy with an active centre (AGN) -was observed in the K band (2 µm) usingthe nucleus to guide the SINFONI AO-Module. The seeing was 0.5 arcsec and thewidth of each slitlet 0.025 arcsec; the totalintegration time on the galaxy was 40 min.At the top is a K-band image of the centralarcsec of the galaxy (left insert) and a K-band spectrum of the nucleus (right). In thelower half are images (left) in the light ofionised hydrogen (Brγ) and molecularhydrogen (H2) lines, together with theircombined rotation curve (middle), as wellas a CO 2-0 bandhead map, which tracescool stars in the nucleus, and images ofbroad and narrow components of the high-excitation [Ca VIII] spectral line (right). Thefalse colours in the images representregions of different surface brightness.

Figure 14: NGC 7469 was observed inK band (2 µm) using the nucleus as the

wavefront reference for the adaptiveoptics and with slitlet widths of 0.025).

At the time of the observations theseeing was 1.1). The total integrationtime on the galaxy is 1 hour 10 min.

Upper left: image of the K-band con-tinuum in the central arcsec of NGC

7469. Upper right: a spectrum extract-ed from the nucleus. Lower left: image

of the molecular hydrogen line togetherwith its rotation curve. Lower centre:

image of the Brγ line. Lower right:image of the CO 2-0 absorption band-

head which traces cool stars.

dimensional distributions of the H2 1–0 S(1)line and the nuclear star forming region,through the CO 2–0 bandhead.

A PA PAIRAIR OFOF SSTTARAR-F-FORMINGORMING

GGALAXIESALAXIES AATT ZZ=2=2It is becoming increasingly clear that mostof the baryonic mass in galaxies was put inplace between about 8 and 11 billion yearsago (redshifts between 1 and 3). However,the real test of our understanding of galaxyformation and evolution is not just when themass accumulated, but rather: How did itaccumulate? Did mass assembly occur in thesame way in different types of galaxies (i.e.,spiral galaxies like the Milky Way and ellip-tical galaxies)? Was the mass accumulationrate mass-dependent so that more massivegalaxies formed earlier in the history of theUniverse? Why do some galaxies have largeamounts of angular momentum, while othersdo not?

To address these questions, astrophysi-cists in Europe and the US have definedsamples of galaxies that are at redshiftswhere most of the mass was likely put inplace, namely around 2. For example,Steidel and collaborators have defined selec-tion criteria based on three optical colourswhich efficiently select actively star-form-ing galaxies at redshifts between 1.5 and 2.5(the so-called “BM and BX galaxies”;Steidel et al. 2004). Galaxies selected thisway are forming stars at approximately20–60 M0/yr, appear to be substantiallyevolved with stellar masses of about1011 M0 and have approximately solarmetallicity. All these properties suggest thatthey are highly evolved systems, eventhough their star-formation rates are sub-stantial. In contrast, they tend to have com-plex UV morphologies and their dynamicalproperties are confusing – typical forgalaxies that are rather unevolved.Interestingly, BM/BX galaxies with elongat-

ed, sometimes flattened morphologies likethose usually associated with discs, appearto have smaller line widths than galaxies ofmore irregular shape. Erb et al. (2004)hypothesized that these galaxies are in thecourse of a merger, and not showing signs ofsystematic rotation which would be associat-ed with a disc.

It is this last argument that led us tocarry out observations with SINFONI of apair of BX galaxies – BX 404/405 – duringcommissioning. Since the morphologies ofBM/BX galaxies are complex and the mostextended structures are not necessarily dueto coherent rotation but to companion merg-ing galaxies, it is difficult or even impossibleto define the “kinematic major axis” (theangle along which the velocity change orgradient reaches its maximum). ExploitingSINFONI's integral field capability, onedoes not a priori have to choose a positionangle, but simply puts the target in the fieldof view and then measures the source kine-matics. If for example a system consists of apair of merging galaxies, then there mighteven be two or more velocity components:Each galaxy could show a rotation curve ofits own inclined by a different angle, andsuperimposed on the relative velocity alongthe line connecting the two galaxies. Withsuch complicated objects, SINFONI is par-ticularly powerful in determining the natureof the sources by obtaining the complete pic-ture of the dynamics at once.

The galaxies BX 404 and 405 are espe-cially interesting because this is an associa-tion of two star-forming galaxies at almostexactly the same redshift only about 3–4 arcseconds apart (about 30 kpc at this redshift).Thus we could cover both objects in a singlepointing. What we found is fascinating: therelative velocity of BX 404 and 405 is onlyabout 150 km/s. They are separated by about30 kpc, while the emission-line object to thesouth and west of BX 404 and 405 has a

velocity of 150 km/s relative to BX405 andabout 300 km/s relative to BX404. The twocomponents of BX405 have velocities thatdiffer by about 70 km/s, with a spatial sepa-ration of about 6 kpc. The velocity and sizeallow us to roughly estimate that BX405 hasa dynamical mass of about 1010 M0.

The overall similar velocities in this sys-tem suggest that this may be an interactingset of galaxies, which could merge to form asingle more massive object. If BX 404 and405 are part of a larger gravitationally boundstructure, their relative velocities and pro-jected distances suggest a total mass ofabout 1011 solar masses. In addition, theratio of [NII]/Hα can be used to give a crudeestimate of the gas-phase metal abundances.In BX 404, the [NII]/Hα line indicates atotal metallicity similar to the Sun. We onlyhave an upper limit for BX405, suggestingthat its metallicity is less than that of BX 404and the Sun.

Obviously, these results are only a firststep: before any robust statement can bemade, a statistically significant sample ofsuch high-redshift galaxies will have to beobserved. Nonetheless, these observationsillustrate the great promise of SINFONI forobtaining a detailed picture of the propertiesof star-forming galaxies in the high-redshiftUniverse.

RREFERENCESEFERENCESH. Bonnet et al. 2003, SPIE 4839, 329F. Eisenhauer et al. 2003, Proc. SPIE 4844, 1548 M. Tecza, et al. 2002, Proc. SPIE, 4842, 36Davies et al. 1998, MNRAS, 293, 189 Davies et al. 2004, ApJ, 602, 148Erb et al. 2004, accepted for publication ApJ

(astro-ph/0404235)Genzel et al. 1995, ApJ, 444, 129Genzel, R. et al. 2003a, ApJ 594, 812Genzel, R. et al. 2003b, Nature 425, 934Ghez, A. et al. 2003, ApJ 586, L127Schödel, R. et al. 2002, Nature 594, 812Steidel, C. C. et al. 2004, ApJ, 604, 534

The Messenger 11724

Figure 15: Image of theGalaxy pair BX 404/405 (red-

shift 2.03) in the light of theHα recombination line (rest-frame wavelength 656 nm).

One of the spectra (upperright) clearly reveals signs of

a velocity shear while theother does not. Such shearmay be a sign of rotation, apossible signature of a disc

galaxy – or of a merger. Therelatively small velocity offsets

of the galaxies suggest thatthey are interacting and may

perhaps merge to form amassive galaxy.