the messenger 122 · 2010. 3. 24. · pickoff module one of the more unusual kmos elements is the...

The MessengerNo. 122 – December 2005

2 The Messenger 122 – December 2005

Surveying the High-Redshift Universe with KMOS

Ray Sharples1

Ralf Bender 2,4

Richard Bennett 3

Keith Burch 3

Paul Carter 3

Mark Casali 6

Paul Clark1

Robert Content1

Richard Davies 4

Roger Davies 5

Marc Dubbeldam1

Gert Finger 6

Reinhard Genzel 4

Reinhold Häfner 2

Achim Hess 2

Markus Kissler-Patig 6

Ken Laidlaw 3

Matt Lehnert 4

Ian Lewis 5

Alan Moorwood 6

Bernard Muschielok 2

Natascha Förster Schreiber 4

Jeff Pirard 6

Suzie Ramsay Howat 3

Phil Rees 3

Josef Richter 2

David Robertson1

Ian Robson 3

Roberto Saglia 2

Matthias Tecza 5

Naranjan Thatte 5

Stephen Todd 3

Michael Wegner 2

1 Department of Physics, University ofDurham, United Kingdom

2 Universitätssternwarte München,Germany

3 UK Astronomy Technology Centre, Royal Observatory, Edinburgh, United Kingdom

4 Max-Planck-Institut für ExtraterrestrischePhysik, Garching, Germany

5 Sub-Department of Astrophysics,University of Oxford, United Kingdom

6 ESO

KMOS is a near-infrared multi-object in-tegral-field spectrometer which hasbeen selected by ESO as one of a suiteof second-generation instruments to be constructed for the VLT. The instru-ment will be built by a consortium of UK and German institutes working inpartnership with ESO and is currently inthe preliminary design phase. KMOS will be capable of obtaining simultane-ous spatially resolved spectroscopy at a sampling of 0.2 arcseconds for upto 24 targets distributed over a field ofview of 7.2 arcminutes diameter.

The past decade has seen remarkableprogress in cosmology, with the combina-tion of measurements from microwavebackground experiments and large-scaleredshift surveys placing precise con-straints on many of the fundamental pa-rameters of the cosmological worldmodel. Photometric selection techniquesand gravitational lensing have opened up the universe beyond z = 1 and allowedthe detection of massive star-forminggalaxies which must have formed within afew billion years of the Big Bang. Theprecise details of the physical processeswhich drive galaxy formation and evolu-tion in these models remain elusive how-ever. To study these processes in detailrequires a capability to map the variationsin star-formation histories, merger ratesand dynamical masses for well-definedsamples of galaxies across a wide rangeof redshifts and environments. Single-inte-gral-field-unit (IFU) spectrographs likeSINFONI/SPIFFI are beginning to provideexquisite views of some of the mostspectacular examples (e.g. Figure 1) butstatistical surveys of these galaxy proper-ties, and follow-up of future surveys of

high-redshift galaxies obtained, for exam-ple, with HAWK-I and SCUBA-2, willrequire a spectrograph that can observemany objects simultaneously. This is the capability which will be delivered by a new instrument now under developmentknown as KMOS (K-band Multi-Ob-ject Spectrograph) which, when commis-sioned on the VLT in 2010, will be uniqueon any 8-metre-class telescope.

For any instrument to address these fun-damental questions about how galaxiesevolve it should: (1) have a substantialmultiplex capability and field of view, com-mensurate with the surface density ofaccessible targets; (2) have the ability toobtain more than just integrated or one-dimensional information since form-ing galaxies are often observed to havecomplex morphologies; (3) be able toresolve the relatively small velocity differ-ences observed in rotation curves, veloci-ty dispersions, and merging galaxy pairs;(4) have the ability to observe severaltargets in proto groups and clusters con-centrated in small areas of the field; (5) enable observations of high-redshiftgalaxies using the well-studied rest-frameoptical diagnostic features used at lowredshift. These general characteristicsimply a near-infrared multi-object spectro-graph using deployable integral field-units(d-IFUs). Deployable IFUs also have asignificant advantage over multi-slit spec-trographs because of the reduced slitcontention in crowded fields and their in-sensitivity to galaxy morphology andorientation. The specific choices made todeliver these capabilities involve a com-plex trade of cost and scope which is reflected in the baseline capabilities ofKMOS listed in Table 1.

Telescopes and Instrumentation

Baseline Design

J = 20 %, H = 30 %, K = 30 %

J = 21.2, H = 21.0, K = 19.2

1.0 to 2.5 µm

R = 3400, 3800, 3800 (J, H, K )

24

2.8 u 2.8 arcseconds

0.2 u 0.2 arcseconds

7.2 arcmin diameter circle

> 3 within 1 arcmin2

edge-to-edge separation of 6 arcsec

Table 1: Baseline design specificationfor the KMOS spectrograph.

Requirement

Instrument throughput

Sensitivity (5 σ, 8 hrs)

Wavelength coverage

Spectral resolution

Number of IFUs

Extent of each IFU

Spatial sampling

Patrol field

Close packing of IFUs

Closest approach of IFUs

3The Messenger 122 – December 2005

2?

0

1000

2000

3000

2.28 2.30 2.32 2.34 2.36 2.38 2.40

6400 6500 6600 6700

0

500

1000

1500

1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4

3000 4000 5000 6000 7000

Hα + [NII]

Observed Wavelength (µm)

Observed Wavelength (µm)

Flux

Den

sity

Flux

Den

sity

Rest Wavelength (Å)

Rest Wavelength (Å)

[OIII][OII]

[SII][SII]

[SII]

[NII]

[NII]

Hα

Figure 1: SPIFFI spectra of the central2.5? of the submillimetre galaxy SMM J14011+0252 showing the oftencomplex morphology of these targets(from Eisenhauer et al. 2003).

KMOS will be mounted on the VLTNasmyth rotator and will use the NasmythA&G facilities. The top-level requirementsare: (i) to support spatially-resolved (3-D)spectroscopy; (ii) to allow multiplexedspectroscopic observations; (iii) to allowobservations across the J, H, and Kinfrared atmospheric windows (extensionto shorter wavelengths may be incor-porated). The baseline design employs 24configurable arms that position foldmirrors at user-specified locations in theNasmyth focal plane, each of which se-lects a sub-field of 2.8 u 2.8 arcseconds.The size of the sub-fields is tailored spe-cifically to the compact sizes of highredshift galaxies. The sub-fields are thenanamorphically magnified onto 24 ad-vanced image slicer IFUs that partitioneach sub-field into 14 identical slices, with14 spatial pixels along each slice. Lightfrom the IFUs is dispersed by three identi-cal cryogenic grating spectrometerswhich generate 14 u 14 spectra, eachwith ~ 1000 Nyquist-sampled spectralresolution elements, for all of the 24 inde-pendent sub-fields. The spectrometerswill each employ a single 2k u 2k Hawaii-2RG HgCdTe detector. The optical layoutfor the whole system has a threefoldsymmetry about the Nasmyth optical axisallowing a staged modular approach to assembly, integration and test. End-to-end raytraces through four of the pick-off arms in one of the three spectrome-ters are shown in Figure 2. Our goal is to

employ careful design choices and ad-vances in technology to ensure thatKMOS achieves a comparable sensitivityto the current generation of single-IFUinfrared spectrometers and gains at leastan order of magnitude in survey speed for typical target fields.

Pickoff module

One of the more unusual KMOS elementsis the pickoff module which relays the lightfrom 24 selected regions distributed with-in the patrol field to an intermediate focusposition at the entrance to the integral-field-unit module. The method adopted forselecting these subfields uses roboticpick-off arms whose pivot points are dis-tributed in a circle around the periphery of the patrol field and which can be drivenin radial and angular motions by twostepper motors which position the pickoffmirrors with a repeatable accuracy of < 0.2 arcsec. The arms patrol in one oftwo layers positioned either side of theNasmyth focal plane to improve the ac-cess to target objects in crowded fields.This focal plane is flattened and madetelecentric by a pair of all-silica field lens-es, one of which forms the entrancewindow to the cryostat. The arm designhas been refined to allow maximum ver-satility in allocation of targets whilstachieving stringent goals on accuracy andreliability. Independent monitoring of armpositions to avoid collisions will be avail-able using both step counting and posi-tion encoders; a collision-detection sensorwill also be implemented as a third level ofprotection. The efficiency of allocation hasbeen benchmarked against several realtarget fields selected from deep imagingsurveys; Figure 3 shows one such config-uration with the arms overlaid on a sam-ple of high-redshift targets selected fromthe FORS Deep Field (Noll et al. 2004).

Figure 2: Optical raytrace through fourpickoff arms, their associated IFUs and one of the spectrometers. Lightexiting the pickoff arms is brought to an intermediate focus using a 3-ele-ment K-mirror, which aligns the edgesof all 24 IFU fields on the sky so thatthey can be put into a compact sparsearray configuration for blind surveys ofcontiguous areas on the sky.


The changing path length within the armis compensated via an optical trombonewhich uses the same lead screw, but witha different pitch, as for the main radialmotion. The pickoff module also containsthe instrument calibration unit and a filterwheel which acts as a focus compen-sation device between the different bands.The cold stop for the instrument is at the base of the arm, after which an inter-mediate image is formed by a K-mirrorassembly which also acts to orientate thepick-off fields so that their edges are par-allel on the sky. A prototype pickoff arm iscurrently being manufactured, which willbe subject to an extensive series of testsin a cryogenic environment before manu-facturing the full batch of 24 arms. A solidmodel of one of the pickoff arms, togetherwith its patrol space envelope, is shown inFigure 4.

Integral Field Unit module

The IFU subsystem contains optics thatcollect the output beams from each of the24 pickoffs and reimages them with ap-propriate anamorphic magnification ontothe image slicers. The anamorphic mag-nification is required in order that the spa-tial sampling pixels (“spaxels”) on the sky are square whilst maintaining Nyquistsampling on the detector in the spectraldimension. The slices from groups of 8 sub-fields are aligned and reformattedinto a single slit for each of the threespectrometers. The optical design of theIFU sub-systems is based on the Ad-vanced Image Slicer concept (Content1997) and draws heavily on experiencedeveloped in building the GNIRS integral-field unit for Gemini South (Dubbeldam etal. 2000). Three off-axis aspheres areused in the fore-optics to facilitate a pro-duction method based on diamond-turn-ing, rather than raster fly-cutting, in orderto improve the surface roughness. Im-portant considerations in developing thedesign for 24 optical trains, have been the need to incorporate manufacturabilityinto the optimisation process, and a de-sire to use monolithic optical componentswherever possible. In the current designthe slicer mirrors are all spherical with thesame radius of curvature, and so are the pupil mirrors. The slit mirrors are toroi-dal with the same radius of curvature inthe spectral direction, but different radii of

Figure 3: I-band image of the FORSDeep Field (Noll et al. 2004) with KMOSpickoff arms assigned to 24 Extreme-ly Red Objects (EROs). The blue armsbelong to the lower plane (and cantherefore be vignetted by arms in theupper plane) whilst the green armspatrol the upper plane. The position-ing efficiency of the arms has beenchecked against a number of importantscience cases which demonstrate ahigh multiplexing factor on interestingtargets.

Figure 4: Solid model of a KMOS armin the upper plane. The lower armscontain identical components but arecompressed vertically. The transpar-ent red region shows the space enve-lope occupied by the arm over its fullrange of motion. Each arm patrolsapproximately 30 % of the pickoff field.

Figure 5: One of the three integratedpickoff and IFU sub-modules showingthe three mounting plates for the pick-off arms, the filter wheels and the IFUoptics. Each sub-module is attached tothe main cryogenic optical bench with-in the cryostat. At the centre of the unit is shown the integrating sphere ofthe calibration unit and the ring mirrorwhich reflects light from the calibrationsources in to the pickoff arms.

Telescopes and Instrumentation Sharples R. et al., Surveying the High-Redshift Universe with KMOS

Milestone

Preliminary Design Review (PDR)

Final Design Review (FDR)

Preliminary Acceptance Europe (PAE)

Preliminary Acceptance Chile (PAC)


Figure 6: Cutaway view of the main KMOS cryostatshowing the entrance window and the pickoff armmodule at the front, and the spectrograph module atthe rear. The cryostat will be an aluminium/stainlesssteel hybrid to reduce weight. Not shown here are theelectronic racks which will mount on a supportingframe around the cryostat.

curvature in the spatial direction. Thisconfiguration was chosen because it is well adapted to the available methodsof machining. Each IFU sub-module pro-duces a 254 mm long slit containing 112 separate slices from 8 subfields. Themechanical design of a single pickoff andIFU sub-module containing eight pick-off arms and eight integral field channelsis shown in Figure 5 and emphasises thethree-fold symmetry of the KMOS sys-tem and the advantages from a mechan-ical perspective of positioning commoncomponents in a single plane.

Spectrograph module

The three identical spectrographs use a single off-axis toroidal mirror to colli-mate the incoming light, which is then dis-persed via a reflection grating and re-focused using a 6-element transmissiveachromatic camera. The gratings aremounted on a 6-position wheel whichallows optimised gratings to be used forthe individual J-, H-, K-bands togetherwith two lower-resolution gratings and theoption of a z-band grating to enhanceversatility (Tecza et al. 2004). Each spec-trograph contains a 2048 u 2048 Hawaii-2RG HgCdTe array which is mounted on a three-axis translation stage in order thatfocus can be adjusted and, if required,some components of flexure can be com-pensated. All three spectrographs aremounted in a plane perpendicular to theNasmyth rotation axis for maximum sta-bility (Figure 6).

Software and electronics

KMOS will be one of the most complexastronomical instruments ever built for a ground-based telescope with over 60 degrees of freedom in the cryogenicmechanisms alone. Robust efficientsoftware and reliable control electronicswill be key to successful long-term op-erations. In addition to instrument controlsoftware and housekeeping diagnostics,KMOS will have an optimised target allo-cation tool, currently known as KARMA, inthe ESO observation preparation soft-ware (P2PP). KARMA will assign arms totargets in a prioritised way, whilst ensur-ing that no invalid arm positions are se-lected and allowing the user to manually

reconfigure the list of allocated targets. A customised data reduction pipeline willbe provided which will allow the observ-er to precisely reacquire the targets duringmultiple visits to the same field and toevaluate the data quality after each read-out. With over 4000 spectra per integra-tion, automatic data processing andreduction methods will be essential to fullyexploit the scientific potential of KMOS.

Project status

KMOS is being built by a balanced con-sortium of UK (University of Durham,University of Oxford and the UK Astrono-my Technology Centre) and German

Date

March 2006

March 2007

March 2010

September 2010

(Universitätssternwarte München and theMax-Planck-Institut für ExtraterrestrischePhysik) institutes working in collaborationwith ESO, who will provide the sciencedetectors and associated readout elec-tronics and software. The project is cur-rently in the preliminary design phase andis expected to be shipped to Paranal in mid-2010. The list of key milestones isgiven in Table 2.

References

Content, R. 1997, Proc. SPIE, 2871, 1295Dubbeldam, M. et al. 2000, Proc. SPIE, 4008, 1181Eisenhauer, F. et al. 2003, The Messenger 113, 17Noll, S. et al. 2004, A&A, 418, 885Tecza, M. et al. 2004, Proc. SPIE, 5492, 1395

Table 2: Key KMOS milestones.


Instrument Concepts for the OWL Telescope


Sandro D’Odorico (ESO)

During the past year ESO has coordi-nated a number of instrument conceptstudies as a complement to the OWLObservatory Design Study. Eight teamsof scientists and engineers from dif-ferent institutes in Europe and ESO haveidentified a variety of science pro-grammes at the frontier of astrophysicsand developed concepts of instrumentsat OWL which would be able to carrythem out. This exercise has provided afirst view of the unique astronomical ob-servations at Blue to IR wavelengthswhich will become possible with a futureEuropean Extremely Large Telescope.

Establishment and overview of the studies

It is generally recognised in the astrono-mical community that it is essential todevelop instrument concepts very early inthe design of a new telescope. Instru-ments represent the vital link between thephoton-collector, however sophisticatedand powerful, and the scientific goals ofthe project. Instrument studies effectivelyprobe the telescope interface and oper-ation scheme and they verify whether therequired scientific observations can beobtained with feasible and affordable in-struments. This path was followed by the VLT project, in which an instrumenta-tion plan was developed almost a decadebefore first light (Reference 1). For OWL,ESO launched eight instrument conceptstudies in 2004 (see Tables 1 and 2) incollaboration with several European insti-tutes.

In the selection of the initial instrumentconcepts, we have been guided by the Science Case for the European ELT (a generic 50–100-m telescope) preparedwithin the OPTICON Network (Editor Isobel Hook) and by preliminary studieson the OWL scientific goals. The selectedinstruments offer various imaging andspectroscopic modes of observing andoperate in different bands from the blue to submillimetre wavelengths. They arewell representative of the different possi-ble modes of operation of OWL andprobe well the telescope’s ultimate capa-bility. The sample is however by no means exhaustive of all potentially unique

observations to be done with an ELT ofthe OWL class. High resolution spec-troscopy in the near infrared and astrome-try at the diffraction limit are two scienti-fically very interesting modes not exploredin this phase.

Six of the studies were led by PIs fromdifferent European Institutes, and twowere coordinated by ESO. The instrumentstudy teams were asked to identify thespecific science drivers and use them todefine the requirements, to develop aninstrument concept and to evaluate itsperformance at OWL. They had to com-pare them with the expected capability ofmajor planned ground-based and space-borne facilities like ALMA and the JWST.They had also to address the dependenceon telescope diameter in the range50–100-m and to underline any criticalaspects in the interface to the telescope.

Critical aspects in cost or the requiredtechnical developments also had to beidentified.

This first effort on possible OWL instru-ments saw the active involvement of morethan 150 astronomers and engineers from over 20 institutes in seven Europeancountries. Through this exercise theybecome familiar with ELT concepts andproduced a first batch of attractive opto-mechanical solutions for the instruments.For most of those who were involved it was a first impact with the “overwhelm-ing” capabilities of a 100-m aperture tele-scope but also with the differences withrespect to the 10-m-class telescopes weare used to work with. All responded in an enthusiastic way to the new challenge.The eight studies were completed in atime frame of 12 months.

Table 1: Instrument Capability and Primary Science Goals.

Instrument

CODEX

QuantEYE

HyTNIC

EPICS

MOMFIS

ONIRICA

T-OWL

SCOWL

Wavelength range

0.4–0.7 µm

0.4–0.8 µm

1.1–1.6 µm

0.6–1.9 µm

0.8–2.5 µm

0.8–2.5 µm

3–24 µm

350–450–850 µm

Main capability

High-velocity-accuracy, visual spectrograph

Photometry at10–3–10–9 second resolution

High-contrast diffraction-limited imaging

Camera-Spectrograph at diffraction limit

Near IR spectroscopy using many deployable IFUs

NIR Camera at diffraction limit

Thermal, Mid-Infrared Imager and Spectrograph

Imaging at sub-millimetre wavelengths

Primary science goals

To measure the dynamics of the Universe

Astrophysical phenomena varying at sub-millisecond time scale

Imaging of massive planets, brightgalactic and extragalactic sources

Imaging and spectroscopy of Earth-like planets

First galaxies in the Universe

Faint stellar and galaxy population

Search, study of planets, high-redshiftHα galaxies

Surveys of dusty regions, of extra-galactic fields for star-forming galaxies

Table 2: Instrument Teams.

Instrument

CODEX

QuantEYE

HyTNIC

EPICS

MOMFIS

ONIRICA

T-OWL

SCOWL

PI

Luca Pasquini

Cesare BarbieriDainis Dravins

Olivier LardièreVirginie BorkowskiAntoine Labeyrie

Norbert HubinMarkus Kasper,Christophe Verinaud

Jean-Gabriel Cuby

Roberto Ragazzoni

Rainer LenzenBernhard Brandl

William Dent

ESO scientist

–

Robert A. E. Fosbury

Guy Monnet

–

Mark Casali

Enrico Marchetti

Hans-Ulrich Käufl

Ralf Siebenmorgen

Institutes

ESO, INAF – Trieste, Observatoire deGenève, IoA Cambridge

University of Padova and Lund University

LISE – Collège de France

ESO and external experts from 10 institutes

LAM, GEPI, CRAL, LESIA, ONERA

INAF – Arcetri, Bologna, Padova, Romaand MPIfA Heidelberg

MPIfA Heidelberg, Leiden University, ASTRON, ESO

UK ATC


Science cases and instrument concepts

Two of the instruments are foreseen forthe Blue-Visual and Red wavelengthbands with natural seeing image quality:CODEX and QuantEYE. They both usethe outstanding collecting power of OWLto do unique science. CODEX makes useof the photon plethora to achieve highS/N ratio, high-resolution spectroscopy offaint stars and quasars with unmatched (~ 1 cm/s) velocity accuracy. The mainscience goal is the direct measurement ofthe dynamics of the Universe, but severalother fields of astrophysics will be boost-ed by CODEX observations, as discussedin Reference 2 and summarised in a sep-arate article on CODEX in this issue of The Messenger (page 10). QuantEYE (Ref-erence 4) explores the temporal dimen-sion of the photon flux. By covering thetime resolution range 10–3–10–9 s, it willpermit for the first time the exploration ofthe quantum properties of the light from avariety of astrophysical sources.

There are two “wide”-field instruments forNIR wavelengths (0.8–2.5 µm): ONIRICA(Reference 7), the imaging camera, and MOMFIS (Reference 6), the multi-fieldspectrograph. Both address many of the“classical” ELT science cases and as suchreveal the power but also the peculiari-ties of observing with OWL. The ONIRICAteam has identified diffraction limitedimaging over a field of ~ 30? as the pri-mary observing mode. Using a MCAOsystem one can expect up to 30 % of thelight to be concentrated in the diffraction

Coma

–0.50

–5

–10

–15

0.5J–K

MK

1.5 210

Fornax

Ages between 10 and 30 MyrAges between 30 and 100 MyrAges between 0.1 and 1 GyrAges older than 2 Gyr

Virgo

ONIRICA 31

30

29

28

0 1 2Exposure time (hours)

3 4

JWST

k(A

B)a

tS

/N=

5

Figure 1a and b: The potential capability of a diffrac-tion-limited NIR camera at OWL for the study of stellar populations in distant galaxies is illustrated inthese figures. On Figure 1a the K limiting magni-tudes for ONIRICA for stellar sources are compared to JWST. The advantage of ONIRICA would be evenlarger in the J-band. With a 60-m or a 30-m tele-scope the curves would have to be lowered by ~ 1and ~ 2.5 mag respectively. Figure 1b shows a theo-retical Colour-Magnitude diagram of an evolved galaxy population in which stars of different ages areidentified. The dashed lines correspond to an mK = 30 star at the distance of different clusters ofgalaxies (see Reference 7 for details).

peak of the PSF over the entire field inperiods of very good natural seeing. Withthis performance the imaging capability of ONIRICA clearly surpasses that of thefuture James Webb Space Telescope(JWST) in terms of limiting magnitudes ofstellar sources. Detailed studies of stellarpopulation in Virgo and up to Coma be-come possible (see Figure 1). In particularwe will be able to investigate for the firsttime the old populations of giant ellipticalgalaxies, a key to understanding the starformation history in the early Universe.

The baseline concept of MOMFIS fore-sees 30 IFU units which can be posi-tioned over the 3;u 3; scientific field ofOWL. Its main scientific goal is spec-troscopy of high-redshift galaxies (z >> 3)to trace the first sources which reionisedthe Universe. Their expected half-lightsizes are typically ~ 0.1? and this valuedrives the IFU size to sub-arcsec and the sampling to 20–30 mas. At a spectralresolution of 4 000, MOMFIS would bemore powerful than JWST in spectrosco-py of faint high-redshift candidates iden-tified by multi-colour JWST imaging. Whileworking far from the diffraction limit,MOMFIS requires a distributed AO system(MOAO) to deliver a moderate concen-tration of light at the sampling resolution.A first run of simulations with naturalguide stars has shown that this might bepossible but with limitations on the skycoverage. The use of laser guide stars ata 100-m telescope has its own problemsand will require further studies.

One of the key science cases for OWL isthe search for Earth-like planets close to nearby stars. Starting from the resultsof the Planet Finder studies for the VLT,the EPICS study (Reference 8) addressesvarious observational approaches to de-tect and characterise Earth- and Jupiter-like planets. Using differential imaging and a coronograph, the EPICS studyinvestigated the possibility to detect theplanets in polarimetry and at the wave-length of biomarkers such as water andOxygen. A NIR IFU and FTS spectros-copy for planet light characterisation arealso briefly discussed. The spectral range in which these modes should oper-ate spans from the Visual to the J-band.EPICS will require a third-generation AO system (XAO) to achieve the diffrac-tion limit with high Strehl and the requiredcontrast with respect to the star light at the wavelengths of operation. A set offirst-order simulations carried out duringthe study suggest that an Earth-like plan-et could indeed be detected with EPICSat a 100-m OWL telescope (see Figure 2).The selection of the final instrument con-


Feedback to the telescope design and future prospects

The studies have identified some criticalaspects of the instrument–telescope interface and pinpointed a number of criti-cal components which will require specialdevelopments. This feedback will betaken into account in the next phase ofthe project.

The present OWL Concept Design under-went an external review at the beginningof November 2005 and the outcometogether with financial and science policyconsiderations by the ESO managementwill be a subject of discussion at the ESO Council in December. If, as we hope,a consensus on the construction of a

European ELT will mature in the nextmonths, the work done on the Instrument Concept Studies will have to be furtherexpanded and tuned to the science goalsand the revised telescope design.

In the meantime, probing of instrumentconcepts is also going on within the ELT Design Study (a network fundedunder the EC FP6 initiative). At the kick-offmeeting of the ELT instrument “SmallStudies” in September 2005, eight instru-ments were identified which will extend or complement the work carried out forOWL. Many of the OWL Instrument Con-cept Study teams are involved in thiseffort and will put to best use their newly-acquired expertise there.

Figure 2 (left): Time to detect an Earth-like planet orbiting a Main-Se-quence star at 5 σ with differentialimaging in water bands in the J re-gion. Assumptions: 80 nm bandwidth, 0.44 atmospheric transmission, 0.32 contrast, t0 = 4 ms, r0 = 20 cm. See Reference 8 for additional details.

Figure 3 (below): Simulations of the re-emitted light at 10 microns, for an un-disturbed circumstellar disc and a discwith a hole of radius 2 AU and 4 AU,respectively, as seen at two different in-clinations and convolved with the OWLPSF (see Reference 3 for more details).The Herbig Ae star is assumed to be at140 parsec.

M2VK2VG2V

40.01

0.1

6 8 10 12 14Distance (Pc)

Exp

osu

retim

e(h

our

s)

16 18 20 22 24 26

1

10

100

undisturbed disc

i=0

deg

i=60

deg

2 AU hole 4 AU hole

figuration, of its primary observing modesand the prediction on its ultimate per-formance will have to wait for more ex-tensive modelling and prototyping in thenext phase of the project.

T-OWL is an imager-spectrograph tooperate in the thermal infrared between 3 and 24 µm. In this spectral region the requirements on the AO system arerelatively modest. A wide range of targetsfrom dusty planetary systems (see Fig-ure 3) to black holes in the nuclei of activegalaxies will be the primary science goalsof T-OWL. In the bands where the at-mosphere is transparent, T-OWL will out-perform the Mid-Infrared Imager MIRI atthe JWST in the observations of point-like sources and offer unique angular res-olution. In spectroscopy, T-OWL can be especially competitive in the observa-tions of narrow emission and absorp-tion features at high resolving power (Ref-erence 3).

SCOWL is a large field (~ 2.5; u 2.5;) sub-millimetre camera to observe in the threesubmillimetre bands at 350, 450 and 850 µm. It capitalises on the expertise ac-quired with the SCUBA1 and SCUBA 2instruments and uses it to draw the con-cept of a powerful survey instrument.SCOWL would supply the ALMA interfer-ometer with a wide range of newly-dis-covered sources for detailed investigation.It benefits from the diffraction limit givenfrom the OWL size (~ 1–2?) at submillime-tre wavelengths without the need of anAO system. The advantage with respectto ALMA in the mapping mode is out-standing as shown in Figure 4. SCOWLrequires the telescope to be installed at a very high and dry site and most likelya tip-tilt mirror correction linked to awater-vapour monitor to achieve its fullpotential (see Reference 5).

HyTNIC explores the application of theconcept of a hyper telescope (Refer-ence 9) as a multi-element imaging inter-ferometer array with a densified pupil to the OWL construction phase. It allowsdirect imaging with enhanced resolutionduring the segment-filling phase of the M1 with a NIR camera and without adap-tive optics, providing observations ofunique scientific value before OWL’s com-pletion.

Telescopes and Instrumentation D’Odorico S., Instrument Concepts for the OWL Telescope


References

The OWL Instrument Concept Studies have beenpublished as ESO internal reports. They can be ob-tained from the PI’s or ESO.

(1) D’Odorico S., Moorwood A. F .M., Beckers, J.1991, Journal of Optics 22, 85

(2) CODEX, Cosmic Dynamics Experiment,OWL–CSR-ESO-00000-0160, October 2005

(3) T-OWL, Thermal Infrared Imager and Spectrographfor OWL, OWL–CSR-ESO-00000-0161, October2005

(4) QuantEYE, OWL–CSR-ESO-00000-0162, October2005

(5) SCOWL, Submillimeter Camera for OWL;OWL–CSR-ESO-00000-0163, September 2005

(6) MOMFIS, Multi Object Multi Field IR Spectrograph,OWL–CSR-ESO-00000-0164, September 2005

(7) ONIRICA, OWL NIR Imaging Camera, OWL–CSR-ESO-00000-0165, October 2005

(8) EPICS, Earth-like Planet Imaging Camera andSpectrograph, OWL–CSR-ESO-00000-0166,October 2005

(9) HyTNIC, Hyper-Telescope Near Infrared Camera,OWL–CSR-ESO-00000-0167, October 2005

The Centre of the Active Galaxy NGC 1097

Near-infrared images of the active galaxy NGC 1097 have been obtained by a team ofastronomers1 using NACO on the VLT. Locatedat a distance of about 45 million light years inthe southern constellation Fornax, NGC 1097 isa relatively bright, barred spiral galaxy seenface-on. It is a very moderate example of an Active Galactic Nucleus (AGN), whose emis-sion is thought to arise from matter (gas and stars) falling into a central black hole. NGC 1097 possesses a comparatively faint nu-cleus only, indicating that the infall rate is small.

The new images probe with unprecedenteddetail the very proximity of the nucleus. The resolution achieved with the images isabout 0.15 arcsecond, corresponding to about 30 light years across. The newly releasedNACO near-infrared images show in additionmore than 300 star-forming regions, a factorfour larger than previously known from HubbleSpace Telescope images. These “H II regions”can be seen as white spots in the image shownhere.

See ESO Press Photo 33/05 for more details.

1 M. Almudena Prieto (Max-Planck Institute für Astro-nomie, Heidelberg, Germany), Witold Maciejewski(University of Oxford, UK), Juha Reunanen (ESO).

850ALMA Compact

450 350 850SCOWL

450 350 λ(µm)

0

0.0001

0.01

100

1000 000

10 000

Figure 4: Relative map-ping speed of SCOWLversus the ALMA Com-pact Configuration.

A colour-composite image of the cen-tral 5 500 light-years wide region of the spiral galaxy NGC 1097, obtainedwith NACO on the VLT. More than 300 star-forming regions – white spotsin the image – are distributed along a ring of dust and gas in the image. Atthe centre of the ring there is a brightcentral source where the active galac-tic nucleus and its supermassive blackhole are located. The image wasconstructed by stacking J- (blue), H-(green), and Ks-band (red) images.North is up and East is to the left. Thefield of view is 24 u 29 arcsec2.


CODEX: Measuring the Expansion of the Universe (and beyond)


Luca Pasquini1

Stefano Cristiani 2

Hans Dekker 1

Martin Haehnelt 3

Paolo Molaro 2

Francesco Pepe 4

Gerardo Avila1

Bernard Delabre1

Sandro D’Odorico1

Jochen Liske1

Peter Shaver1

Piercarlo Bonifacio 2

Stefano Borgani 2

Valentina D’Odorico 2

Eros Vanzella 2

François Bouchy 4

Miroslava Dessauges-Lavadsky 4

Cristoph Lovis 4

Michel Mayor 4

Didier Queloz 4

Stéphane Udry 4

Michael Murphy 3

Matteo Viel 3

Andrea Grazian 5

Sergei Levshakov 6

Lauro Moscardini 7

Tommy Wiklind 8

Shay Zucker 9

1 ESO2 INAF – Osservatorio Astronomico

di Trieste, Italy3 Institute of Astronomy, Cambridge

University, United Kingdom4 Observatoire de Genève, Switzerland5 INAF – Osservatorio Astronomico

di Roma, Italy6 Ioffe Physical-Technical Institute,

St. Petersburg, Russian Federation 7 University of Bologna, Italy 8 ESA-STScI9 Weizmann Institute of Science, Tel Aviv,

Israel

CODEX, a high-resolution, super-stablespectrograph to be fed by OWL, the most powerful telescope ever con-ceived, will for the first time provide thepossibility of directly measuring thechange of the expansion rate of the Uni-verse with time … and much more.The concept of this instrument has beenstudied in a collaboration launched byESO in July 2004 together with INAF –Osservatorio Astronomico di Trieste,

the Institute of Astronomy, CambridgeUniversity, and the Observatoire deGenève.*

The expansion of the Universe

The discovery of the expansion of theUniverse in the late 1920s by EdwinHubble was a major milestone in cosmol-ogy. It brought to an end the belief heldby most physicists of the time includingAlbert Einstein that the Universe is staticand not evolving. Hubble’s discovery ofthe expansion of the Universe has beenconfirmed by a vast range of astronomicalobservations. It eventually led to the wide-ly accepted Hot Big Bang Theory, whichpredicts that the Universe was very denseand hot at early times, and is an essen-tial aspect of the cosmological standardmodel.

Einstein’s theory of General Relativity (GR)led to the description of the Universe as ahomogeneous and isotropic four-dimen-sional space time, the so-called Friedman-Robertson-Walker (FRW) Universe. In1922 Friedman (and independentlyLemaître in 1927) found that Einstein’soriginal equations did not allow staticsolutions. Ironically this prompted Einsteinto “spoil” his new theory of gravity byintroducing a cosmological constant toallow static solutions in his field equa-tions. This same cosmological constant,which Einstein in his later years consid-ered as “the biggest blunder” of his life, has now become another pillar of themodern cosmological standard model.

In the Early Universe a large vacuum ener-gy density acting like a cosmological con-stant is believed to have been responsiblefor the rapid expansion of the Universe ina phase called Inflation. Furthermore,recent observations of the luminosity dis-tance of Type I a supernovae have estab-lished the presence of a form of DarkEnergy which appears to have an effectsimilar to that of Einstein’s cosmologicalconstant within the framework of a FRWUniverse.

It is possible in principle to directly meas-ure the change of the expansion rate

* See also the article by Sandro D’Odorico on page 6in this issue of The Messenger.

of the Universe with time. If we consider in an expanding FRW Universe the light of a source which is emitted at time teand received at time t0, the change ofredshift z of the source with time can beexpressed as: z

.= (1 + z)H0 – H(te).

The time derivative of the redshift of lightemitted by a source at fixed coordinatedistance is thus related in a simple man-ner to the evolution of the Hubble param-eter H(te) between the epoch of emissionand reception. The Hubble parameter isrelated to the energy content of the Uni-verse as

where Ωtot = Ωmat + ΩR + Ωde and Ωmat,ΩR, Ωde are the energy density of matter,radiation, and dark energy expressed interms of the critical density, respectively,and a and a0 are the scale factor at thetime te and t0. The dark energy density ischaracterised by an equation of state of the form pde = wρdec

2, and w = –1 cor-responds to the case of a cosmologicalconstant. Note that we do not know muchabout the dark energy term, and its red-shift dependence could well be morecomplicated than parameterised here by a simple equation of state. At the red-shifts here considered (z ~ 2–5) the radia-tion energy density is small ΩR << Ωtotand can be neglected. The majority of as-tronomical observations, most prominent-ly those of the CMB, supernovae, Lyαforest and the clustering of galaxies areconsistent with a FRW Universe with no curvature and a cosmological constantwhich corresponds to an energy densityabout twice that of the matter at present.

It is important to characterise the physicaleffects of “dark energy” as completely aspossible and in particular it is essential toestablish whether the dark energy actuallyhas the dynamical effects expected in GR.We recall that all observational constraintsare basically geometric in nature as theymainly constrain the angular diameter dis-tance to the last scattering surface (CMB)and the luminosity distance at moderateredshifts (supernovae). The constraints onactual dynamical effects of the cosmo-logical constant as probed by the cluster-ing of the matter distribution are coupledin a complicated way to geometrical con-straints and are actually rather weak.

H = H0 Ωmat

a0 +a

3

(1– Ωtot )a0

a

2 1/2

+4

+Ωde

a0

a

3 (1+w )

ΩR

a0

a


=10 –10z

=10 –11

z = 5 u10 –10.

ΩΛ = 0.0

ΩΛ = 0.4

ΩΛ = 0.6

ΩΛ = 0.7

ΩΛ = 0.8

ΩΛ = 0.9

Redshift86420

10

–1–

2–

3

.

.z.

v(c

ms–1

yr–1

)

Figure 1 (left): Evolution of v.

as a func-tion of redshift. The cosmologicalparameters have been fixed to Ωtot = 1,H0 = 70 and different values of ΩΛhave been considered. The ΩΛ = 0.7,ΩM = 0.3 cosmology is shown with a blue solid line, and the Einstein-de Sitter model is plotted with a redlong-dashed line. The filled circlesconnected by dotted lines show loci ofconstant z

., in units of yr –1.

The CODEX experiment is conceptuallyvery simple: by making observations of high redshift objects with a time intervalof several years, we want to detect and use the wavelength shifts of spectralfeatures of light emitted at high redshift to probe the evolution of the expansion ofthe Universe directly.

It is convenient to express the expectedwavelength shift over a period ∆t in termsof the velocity of the equivalent Dopplershift, v = (∆λ /λ)c. Figure 1 shows the ex-pected change of Doppler shift for a rangeof FRW models with no curvature as afunction of redshift. There are a few thingsto note. The wavelength shift has a very characteristic redshift dependence.At some redshifts the wavelengths are“stretched” while in others they are “compressed”. The wavelength shift cor-responds to a Doppler shift of about1–10 cm/s over a period of 10 years.

This amount is extremely small, andbrought Sandage (1962) to conclude thatsuch a measurement was beyond ourcapabilities. Why do we think that this ex-periment is possible now? For threereasons: (1) Extremely Large Telescopessuch as OWL should be capable of pro-viding the huge number of photonsrequired. (2) In the last two decades ourcapability of accurately measuring wave-length shifts of astronomical sources hasdramatically improved (Mayor et al. 2002).(3) A suitable class of astronomical ob-jects for this measurement has been iden-tified: the Lyα forest lines, which areextremely numerous and beautifully tracethe cosmic expansion with negligiblepeculiar motions (at least ten times small-er than the Hubble flow).

Measuring the wavelength shift in high redshift objects

A priori it is not obvious which objectsand which spectral features are best suit-ed for a precise measurement of z

.. For

a given energy flux the precision of thefinal measurements will increase with thesharpness of spectral features (less noise) and increasing wavelength (morephotons). Another important considera-tion is the expected peculiar accelerationassociated with peculiar motions relativeto the Hubble flow, which will act as ad-

ditional noise. The numerous absorptionlines in the spectra of high-redshift QSOs,which make up the so-called Lyα forest,appear to be the most promising targetsfor a measurement of z

.. The most striking

feature of the Lyα absorption is the largenumber of lines in a single spectrum.

There are about one hundred suitable fea-tures in a single spectrum which have a typical width ∆λ line of 30 km/s. WithQSO absorption spectra we can probe awide redshift range from z ~ 1.5 up to 4and beyond. There is a generally accept-ed paradigm for the origin of the Lyα for-est absorption and the associated metalabsorption which has been established by extensive comparison of cosmologicalhydrodynamical simulations and analy-tical calculations with the exquisite data

which have become available from 10-m-class telescopes (see Rauch (1998)for a review). The peculiar accelerationdistribution as derived from state-of-the-art hydrodynamical simulations showsthat for the vast majority of these systemsthe expected acceleration is indeed neg-ligible. The Lyman series absorption linesare thus ideally suited to measure theglobal evolution of the Hubble parameter.Figure 2 illustrates the Lyα absorbers andthe cosmic shift with time.

The simulations

In order to quantitatively assess the fea-sibility of the measurement, Monte-Carlosimulations have been carried out inde-pendently by several groups. The high-re-

4980 4990 5000 5010 5020

00.

51

Wavelength (Å)

Flux

50304970

Figure 2 (below): This figure shows thedisplacement of intergalactic absorp-tion features caused by the evolution of the Universe’s expansion rate. Weshow a small portion of the simulatedLyα forest in the spectrum of a z ~ 3QSO as we would observe it today(blue line) and 107 years from now (redline). The time interval between the two observations is huge (107 years) forthe sake of visualisation. In the realexperiment it will be about six orders ofmagnitude smaller, as will be thedisplacement of the spectral features.


Telescopes and Instrumentation Pasquini L. et al., CODEX: Measuring the Expansion of the Universe

solution spectra of QSO were simulated,noise added and the process repeated for the second epoch. The pairs of spec-tra so produced were compared and the‘measurement’ performed. Figure 3 showsthe difference expected among pairs ofspectra, where the second epoch spec-trum has been redshifted according to dif-ferent cosmological models. The results ofthe simulations agree with the fact that, if the lines are resolved, the final accuracyof the experiment can be expressed as afunction of signal-to-noise (S/N) ratio andredshift as:

(where S/N refers to a pixel of 0.0125 Å).This implies that observing each ep-och for (e.g.) 40 QSOs with a S/N ratio of2 000 each, an accuracy of 1.5 cm/seccan be obtained. Figure 4 shows that fora QSO of magnitude 16.5 and 2 000 hoursof observations, such a S/N ratio is withinreach for some of the larger next-genera-tion telescopes currently under considera-tion, provided that the whole system hasan efficiency comparable to that of UVESat the VLT.

A sufficient number of bright QSOs is al-ready available. Selecting objects frompublished catalogues, we find 91 QSOsbrighter than m = 16.5, out of which 25 have redshifts between 2 and 4, and the number of suitable objects shouldincrease with the large, all-sky photomet-ric surveys planned in the coming years.

The instrument concept

We have therefore developed an instru-ment design concept with the character-istics given in Table 1. In order to obtain a resolving power of 150 000 on a seeing-limited ELT (1 arcsecond aperture on a 60-m telescope or 0.65 arcsecond on a 100-m), five identical spectrographs areforeseen. To obtain the highest stability,each spectrograph will be contained in avacuum tank, hosted in a temperature-stabilised room nested in an environmen-tally quiet laboratory. In order to keep to alimited size for each spectrograph, severalnew concepts (pupil anamorphysm andslicing, special crossdisperser) have beenadopted, and each spectrograph will havean echelle grating only twice the size

σv = 1.4 cm/s

S/N2350

–1 NQSO

30

–1/2

1+ ZQSO

5

–1.8

Figure 3: Flux difference of two simulat-ed, noiseless Lyα forest spectra at z = 3.1 taken ∆t = 10 years apart fordifferent cosmological parameters as indicated and H0 = 70 km/s/Mpc.

Figure 4: Telescope + CODEX efficien-cy versus telescope diameter, for a S/N ratio of 12 000, integration time of2 000 hours and a QSO magnitude of V = 16.5. The plot shows the totalefficiency that the system should have to reach the goal with the aboveassumptions. The red line is the UVES+ VLT efficiency.

ΩM, ΩΛ = 1.0, 0.0ΩM, ΩΛ = 0.3, 0.0ΩM, ΩΛ = 0.3, 0.7

4960 4980 5000 5020 5040Wavelength (Å)

–5

u10

–6

5u

10–

610

–5

Ep

och

1–E

poc

h2

0

20406080100Telescope Diameter (m)

VLT + UVESMaximum Efficiency

00.

51

1.5

Tota

lEffi

cien

cy

Assumed: V = 16.5, pixel = 0.015 Å, total S/N = 12 00036 QSOs each with S/N = 2 000/pixel in 2 000 hours

2

of those of UVES, with an 8 k u 8 k de-tector.

While it is possible to predict the behav-iour of the individual elements of CODEX, it will be extremely difficult to model thewhole system, including, for instance, thecomplex interactions with the telescope.We therefore anticipate that a full CODEXunit, exactly similar to one of the five in-stalled at OWL, has to be developed andoperated for several years at the VLT, inorder to gain the experience and to (im)-prove the CODEX concept.

Chasing the systematics

When aiming at precise measurementswhich go almost a factor 100 beyondpresently achievable performance, special

care must be taken to account for subtlesystematic effects. “Local” forms of noiseare relevant; an evaluation of the barycen-tric correction terms is given in Table 2,which shows that the corrective terms areunder control and within reach at present.One important number missing from Table 2 is the value of the acceleration ofthe Sun in the Galaxy, which is compa-rable to the amount of the cosmic signal.This important term can in principle be measured by CODEX observing QSOswell distributed in the sky, but it will be determined with superior accuracy bythe ESA GAIA satellite at a level of 0.5 mm/sec/year – an accuracy ten timessmaller than the cosmic signal.

The accuracy of the wavelength calibra-tion is another concern (we shall recallthat a shift of 1 cm/sec corresponds to


Table 1: Main characteristics of CODEX.

Acceptance aperture on the sky

Location

Feed

Peak DQE including injection losses(with GLAO)

Number of unit spectrographs

Unit Spectrograph dimensions

Spectral resolution

Wavelength coverage

Spectrograph layout

Echelle

Crossdisperser

Camera

Detector

Noise performance

Sampling

1 arcsec for 60 m, 0.65 arcsec for 100 m

Underground in nested thermally stabilised environment

Coudé feed

14 % (Coudé feed)

5 (11 for 100-m OWL and 1 arcsec aperture)

Diameter 2.4 u 4 m (vacuum vessel)

150 000

446–671 nm in 35 orders

White pupil

41.6 l/mm, R4, 170 u 20 cm, 4 u 1 mosaic

VPHG 1500 l/mm operated off-Littrow

Dioptric F/2.3, image quality 30 µm

CCD mosaic 8 k u 8 k, 15 µm pixelsStabilised to a few mK

Photon shot noise limited for mv = 16.5 in 10 minutes

4 pixels per FWHM

Table 2: Sensitivity matrix of the accuracy of the barycentric correction with regard to their input parameters.

Parameter

Earth orbital velocity– Solar-system ephemerides

Earth rotation– Geoid shape– Observatory coordinates– Observatory altitude– Precession/nutation corrections

Target coordinates– RA and DEC– Proper motion– Parallax

Relativistic corrections– Local gravitational potential

Timing– Flux-weighted date of observation

Induced error on the correction [cm s –1]

< 0.1

~ 0.5< 0.1< 0.1< 0.1

?~ 0~ 0

< 0.1

?

Comment

JPL DE405

Any location in atm.along photon path may be chosen

70 mas 1 cm s–1

negligiblenegligible

0.6 s 1 cm s–1

about 3 Å shift on the detector). Testsmade with HARPS indicate that the Th-Arlamps used in most spectrographs arenot adequate for such accuracy, and thata new calibration system should be de-vised. In addition, such a novel calibrationsystem must be perfectly reproducibleand stable over the long term (20 years ormore) of the duration of the experiment.We have been investigating with the Max-Planck-Institut für Quantenoptik (MPQ)the possibility of using new superb stan-dards based on newly-developed la-ser frequency combs. The group at MPQis led by Prof. Hänsch, one of the NobelPrize Winners in Physics this year.

CODEX beyond the measurement of the cosmic signal

The scientific applications of CODEX, as a high-resolution spectrograph withextremely high performance fed by OWL,will go well beyond the main experimentproposed above. In the following we give a glimpse of three outstanding appli-cations:

Search for variability of fundamentalconstants

Fundamental constants are supposedlyuniversal and unvarying quantities. Any measured variations would have far-

reaching consequences for the unifiedtheories of fundamental interactions, for the existence of extra dimensions ofspace and/or time and for the exist-ence of scalar fields acting in the late uni-verse. Only astronomical observationshold the potential to probe the values offundamental constants in the past, and in remote regions of space. In 2001,observations of QSO absorption linesbrought the first hints that the value of thefine-structure constant α – the centralparameter in electromagnetism – mightchange over time (Murphy et al. 2001),but more recent observations are consis-tent with a null result. An effective two to three order of magnitude precision gainis foreseen with a spectrograph with R ≈ 150 000 at OWL. The accuracy of the∆α/α variation measurements will be a few times 10–9 – more precise than anyother astronomical and geological meas-urement.

Search for other earths

Exo-planets and, in particular, terrestrialplanets in habitable zones, will be one of the main scientific topics of the nextdecades, and one of the main OWL sci-ence drivers. CODEX with OWL will leadthe discoveries in at least three maincases in exo-planetary science, providingwith unique capabilities and observations:(i) discovery and confirmation of rockyplanets, (ii) search for long-period planets,(iii) Jupiter-mass planets around faint stars.

The need for a ground-based follow-upfacility capable of high radial velocity ac-curacy has been stressed in the recentESO-ESA working group report on solarplanets: a high-precision radial-velocityinstrumentation for the follow-up of astro-metric and transit detections, to ensurethe detection of a planet by a second in-dependent method, and to determine its true mass. For Jupiter-mass planets,existing instrumentation may be techni-cally adequate; for Earth-mass candi-dates, special-purpose instrumentation(like CODEX) on a large telescope wouldbe required. CODEX will also allow us to search for hot Jupiters around solar-mass stars in different environments andstar-forming histories, such as globularclusters and nearby companions to theGalaxy.

N

E


Afterglows of Elusive Short Gamma-Ray Bursts

An international team of astronomers1 hasfor the first time observed the visible light from a short gamma-ray burst (GRB).Using the 1.5-m Danish telescope at La Silla (Chile), they showed that theseshort, intense bursts of gamma-ray emis-sion most likely originate from the violentcollision of two merging neutron stars.The same team has also used the VLT toconstrain the birthplace of the first evershort burst whose position could be pin-pointed with high precision. The resultswere published in the October 6 issue ofthe journal Nature.

Gamma-ray bursts, the most powerfultype of explosion known in the Universe,have been a mystery for three decades.They come in two different types, longand short. Over the past few years, inter-national efforts have convincingly shown

that long gamma-ray bursts (longer thanabout 2 sec) are linked with the explosionof massive stars. It is thought that short-duration GRBs may be due to themerging of two neutron stars; they have evaded optical detection for morethan 30 years.

In the night of July 9 to 10, 2005, theNASA HETE-2 satellite detected a burstof only 70-millisecond duration and,based on the detection of X-rays, wasable to determine its position in the sky. Thirty-three hours after, Jens Hjorthand his team obtained images of thisregion of the sky using the Danish 1.5-m

1 Jens Hjorth (Dark Cosmology Centre – DARK, NielsBohr Institute, University of Copenhagen), DarachWatson (DARK), Johan P. U. Fynbo (DARK), Paul A. Price (Institute for Astronomy, University ofHawaii), Brian L. Jensen (DARK), Uffe G. Joergensen(DARK), Daniel Kubas (ESO), Javier Gorosabel(Instituto de Astrofísica de Andalucía), PàllJakobsson (DARK), Jesper Sollerman (DARK andDepartment of Astronomy, Stockholm University),Kristian Pedersen (DARK), and Chryssa Kouveliotou(NASA/Marshall Space Flight Center). The team is part of the Gamma-Ray Burst Afterglow Collabo-ration at ESO (GRACE) carrying out gamma-rayburst afterglow studies.


First image in the visible (more precisely, in the so-called R-band) of a short gamma-ray burst. The imagewas taken with the Danish 1.5-m telescope and theDFOSC camera at La Silla on July 11, 2005. It showsthe gamma-ray burst to be situated on the edge of alow-redshift galaxy.

telescope at ESO La Silla. The imagesshowed the presence of a fading source,sitting on the edge of a galaxy.

The burst resides 11000 light years fromthe centre of a star-forming dwarf galaxythat is about 2 400 million light years awayand is quite young – about 400 millionyears old. From observations conducteduntil 20 days after the burst, the as-tronomers ruled out the occurrence of anenergetic hypernova as found in mostlong GRBs. This supports the hypothesisthat short GRBs are the consequence of the merging of two very compact stars.

(Based on ESO Press Release 26/05)

Primordial nucleosynthesis

Standard Big bang nucleosynthesis pre-sents a pressing cosmological conun-drum. There is some evidence suggestinga cosmological origin for 6Li, and the stel-lar value for primordial 7Li does not agreewith primordial Deuterium from QSOs andwith Ωb from WMAP. Although Li obser-vations in low metallicity Galactic halostars are plagued by possible systematicuncertainties due to modelling of stellaratmospheres and the treatment of con-vection, it is possible that both discrepan-cies can be reconciled with physics be-yond the standard model during theQuark-Hadron phase. CODEX will provide

the first observations of 7Li and 6Li indwarf stars in galaxies of the Local Groupand will make it possible for the first timeto measure the interstellar 7Li/6Li ratio in unprocessed material of High-VelocityClouds. The latter is a direct and robustprobe of the yields of the Big Bang Nucle-osynthesis yields, providing importantinsights on whether new physics is play-ing a role in the early Universe.

In addition to these selected applications,many other science cases have emergedover recent years. Stellar oscillations, the study of the most metal-poor stars inour Galaxy and in its local group compan-ions, the use of cosmochronometers to

determine the age of the Universe, thehistory of the metal enrichment of the Uni-verse … all these cases (and surely manymore) will receive a major boost fromCODEX on OWL, far beyond what can benow conceived.

References

Mayor, M. et al. 2002, The Messenger 110, 9Murphy, M. T. et al. 2001, MNRAS 327, 1244Rauch, M. 1998, ARA&A, 36, 267Sandage, A. 1962, ApJ 136, 319



ALMA News

Tom Wilson (ESO)

ALMA Cost Review

An independent international Cost Reviewof the ALMA project was held in Gar-misch-Partenkirchen on October 13–16.The general response of the Cost Re-view Committee was positive. The threeALMA Project Managers (Tony Beasley,Joint ALMA Office; Adrian Russell, NorthAmerica; Hans Rykaczewski, Europe)wrote that “The response from the com-mittee is pretty much as good as it couldbe ...” Massimo Tarenghi, the ALMADirector, added “This excellent outcomereflects the very hard work of manypeople in the ALMA project, and the con-scientious way in which the review wascarried out.” More details of the review willfollow when the final report becomesavailable.

Milestones

There are a number of milestones for thecompletion of ALMA. The largest is thecontract for antennas, since this contractis more than 30 % of the total cost of the bilateral (North America and Europe)ALMA. Here, we report the news on an-tenna procurement.

The ESO Council resolution from themeeting of September 30 states that “theALMA project is affordable and com-patible with ESO’s strategic priorities”. Itfurther states that the ESO Council“requests the Finance Committee to pro-ceed to decide on the proposal to awarda contract for the production of the ESO ALMA antennas”. The ESO FinanceCommittee met on October 5, and hasapproved the negotiation and conclusionof a contract with the selected vendor. At the ALMA Board meeting in Santiago,Chile, on November 1, it was announcedthat the ALMA Board concurs with theALMA Director’s recommendation that theEuropean Executive proceed with the is-suance of a contract to procure its shareof the ALMA antennas.

Another important milestone is the FrontEnd Integration Center. The Band 3 (3 mil-limetres, or 84–116 GHz) and Band 6 (1.3 mm, or 211–275 GHz) Front Ends are

built in North America, while Band 7 (0.9 mm, or 275–373 GHz) is built at IRAMGrenoble and Band 9 is built at SRON,Groningen, Netherlands (0.5 mm, or602–720 GHz). The decision has beenmade to have two Integration Centres,one in Europe at the Rutherford-AppletonLaboratory in Oxfordshire, UK, the oth-er in North America, at the NRAO in Char-lottesville.

With this level of progress, we are nowmuch closer to the beginning of theassembly of individual components toproduce a working system.

ALMA software development

It is expected that the ALMA array will beused by a relatively large community ofastronomers that are less experienced inusing radio interferometers and/or do-ing science at millimetre and sub-millime-tre wavelengths. Therefore, it is of utmostimportance to take this into account indevelopment of software products that will eventually be used by the community.

Observing preparation

The ALMA Observing Tool (OT) will be thesoftware tool that supports astronomersin constructing a full Observing Project forthe ALMA Observatory. Basically, suchObserving Programmes will be submittedto the Observatory in two parts. The

first is a Phase I Observing Proposal thatwill have its emphasis on the scientificjustification of the proposed observations.The second part of the project is thePhase II Observing Programme that canbe submitted to the ALMA Observatory if observing time has been granted by the Time Allocation Committee (TAC) on the basis of the accepted proposal.

Central in the OT is the creating of a setof Scheduling Blocks (SBs) which are re-quired to drive observing with ALMA. The SB is the smallest (indivisible) unit inALMA observing that can be scheduledindependently. It is self contained andusually provides scientifically meaningfuldata. The SB contains a full description of how the science target and the calibra-tion targets are to be observed, and sets of SBs can be combined with a de-scription for the post processing of the data, ultimately resulting in an image.

In order to serve both less experiencedand experienced astronomers, the OT willbe equipped with two so-called “Views” to make Observing Programme prepara-tions. It is intended that the main view on the ALMA system will be the “ScienceView”. As the name indicates, in this viewthe users can concentrate on inputting the science requirements of their observ-ing programme: the area to be observedfor each target, required sensitivity andfrequencies. For most observing even ex-perienced users should only need to usethis view. The required SBs will be con-

A view of a presentationon the first day of theindependent ALMA CostReview held atGarmisch-Partenkirchen.The total number ofattendees was about 80.The chair was StevenBeckwith (STScI) and thevice chair was Thijs de Graauw (SRON).


structed by the system and the user willonly be bothered with system parame-ters when this is absolutely necessary, ingeneral detailed parameters will be deter-mined from the science input.

However, it is recognised that for someprogrammes, and indeed for develop-ing new observing modes, an ALMA-ex-perienced observer will need more. The“System View” (or expert mode) willprovide such a user with a complete setof parameter fields that enable a detailedspecification of each scheduling block:the observing process of science and cal-ibration targets, including data acquisi-tion and reduction recipes. These param-eters include the setting of the localoscillator, the upper and lower side bands,the correlator parameters and the selec-tion of the basebands and subband setswithin each baseband. Whichever “View”is used, SBs must be created.

An interesting functionality in the SystemView that has been added recently is the Visual Spectral Editor. This will helpsomewhat experienced users to care-fully position LO frequency, and the fre-quency characteristics of the basebandsconfiguration within the ALMA receiverband. In the figure a part of this editor isvisible. It shows the upper part of theVisual Editor display with the all ALMAFrequency bands (note that these will not all be available; see also http://www.eso.org /alma/specifications /FreqBands.html ), the transmission curveof the atmosphere over the full ALMAobservable spectrum (note Band 6 hasbeen selected), the setting to the LO frequency and upper and lower sidebands, and one baseband selection. A later version of this tool will aid less ex-perienced observers, without subjectingthem to the detailed setup.

The development of the Observing Tool isa shared effort between the UK Astron-omy Technology Centre in Edinburgh,ESO and the NAOJ, with science adviceprovided by Osservatorio Astrofisico di Arcetri and the NAOJ. Development iscurrently focused on the Science andSystem View of the Phase II ObservingProgramme definition. Work has been on-going for two years with a major and aminor release each year. User test cyclesimmediately follow releases of the soft-

ware. Currently, it is planned to have fourmajor user test sessions and three minorones. The major user tests provide impor-tant and timely input feedback to theteam for possible upgrades in the follow-ing releases and are performed by typical-ly eight people. Minor test sessions arefollow-ups of the major releases but pos-sibly include further tests and input for thedevelopment team toward the next majorrelease. The next major user test will takeplace this November. It is planned to holdwider “beta testing” in advance of the firstrelease of the tool to the community.

Data reduction

All ALMA data will be reduced using theALMA offline reduction and imagingpackage. This package is based on theC++ code base in AIPS++ but the code it-self is undergoing some fairly majorchanges to optimise it for ALMA and theuser interface is being redesigned. Formany observations the automated calibra-tion and imaging pipelines will producereference images suitable for analysis.

During the past 1.5 years, ALMA has con-ducted external user tests every sixmonths to ensure that the reduction soft-ware development is adequate for ALMAneeds. This is an ongoing process de-signed to incrementally test functionalityas it is developed. To date, three testcycles have been completed. The testinghas focused on verifying that the under-lying C++ code has adequate functionality

and is robust enough for ALMA needs(e.g. make sure that users can processdatasets from end-to-end using specificALMA cases). The user interface (both the GUIs and the script language) willchange significantly over the next severalyears and thus, current testing could notyet evaluate the robustness or user-friendliness of the interface. Essentially, if testers could do what was necessary to get a scientifically accurate image andevaluate the image quality (even if thesyntax or process was a bit complicated)then the test was considered a success.User interface elements will begin to betested in early 2006. External testers havebeen volunteers from the world-wideastronomical community. So far, onlyexpert interferometrists have been askedto test the software. Later, when the new user interface is developed, noviceusers may be asked to help with the test-ing. The tests have been successful to date – all testers have been able to fill,edit, calibrate, image, and analyse thetest data sets. Based on these tests, itappears that the offline reduction andimaging package is on schedule for meet-ing ALMA data processing needs at the beginning of early science operations. For more details, see the AIPS++ home-page (http://aips2.nrao.edu/docs/aips++.html ) or the latest test report(http://aips2.nrao.edu/projectoffice /almatst2.0/Offline.Test3.Report.7july05-final.pdf ). (Based on a contribution byRein Warmels (ESO) and Debra Shepherd(NRAO).)

A typical display produced by theObserving Tool (OT) using the VisualSpectral Editor. The graphics showsthe ALMA receiver bands and the user-selected positions for the basebandsand sidebands. Also, the atmospherictransmission curve is displayed.

Telescopes and Instrumentation Wilson T., ALMA News



ALMA Antenna Contract Signed

On December 6, ESO signed a contractwith the consortium led by Alcatel AleniaSpace and composed also of EuropeanIndustrial Engineering (Italy) and MT Aero-space (Germany), to supply 25 antennasfor the Atacama Large Millimeter Array(ALMA) project, along with an option foranother seven antennas. The contract,worth 147 million euros, covers the design,manufacture, transport and on-site in-tegration of the antennas. It is the largestcontract ever signed in ground-basedastronomy in Europe.

The ALMA antennas present difficult tech-nical challenges, since the antenna sur-face accuracy must be within 25 microns,the pointing accuracy within 0.6 arc sec-onds, and the antennas must be able tobe moved between various stations onthe ALMA site. This is especially remark-able since the antennas will be locatedoutdoor in all weather conditions, withoutany protection. Moreover, the ALMA an-tennas can be pointed directly at the Sun.ALMA will have a collecting area of morethan 5 600 square metres, allowing for unprecedented measurements of extreme-ly faint objects.

The signing ceremony took place on De-cember 6, 2005 at ESO Headquarters in Garching, Germany. “This contract rep-resents a major milestone. It allows us to move forward, together with our Ameri-can and Japanese colleagues, in this very ambitious and unique project”, saidESO’s Director General, Dr. CatherineCesarsky. “By building ALMA, we are giv-ing European astronomers access to the world’s leading submillimetre facility atthe beginning of the next decade, there-by fulfilling Europe’s desire to play a majorrole in this field of fundamental research.”

Pascale Sourisse, Chairman and CEO ofAlcatel Alenia Space, said: “We would liketo thank ESO for trusting us to take onthis new challenge. We are bringing to thetable not only our recognised expertise in antenna development, but also ourlong-standing experience in coordinatingconsortiums in charge of complex, high-performance ground systems.”

ALMA is an international astronomy facili-ty. It is a partnership between Europe,North America and Japan, in cooperationwith the Republic of Chile. The European

contribution is funded by ESO and Spain,with the construction and operationsbeing managed by ESO. A matching con-tribution is being made by the USA andCanada, which will also provide 25 anten-nas. Japan will provide additional an-tennas, thus making this a truly worldwideendeavour.

ALMA will be located on the 5 000 m highLlano de Chajnantor site in the AtacamaDesert of Northern Chile. ALMA willconsist of a giant array of 12-m antennasseparated by baselines of up to 18 kmand is expected to start partial operationby 2010–2011.

The excellent site, the most sensitive re-ceivers developed so far, and the largenumber of antennas will allow ALMA tohave a sensitivity that is many times betterthan any other comparable instrument.“ALMA will bring to sub-millimetre astron-

omy the aperture synthesis techniques ofradio astronomy, enabling precision imag-ing to be done on sub-arcsecond angularscales, and will nicely complement theESO VLT/VLTI observatory”, said Dr. HansRykaczewski, the ALMA European ProjectManager.

A prototype antenna has already beenbuilt by Alcatel Alenia Space and Euro-pean Industrial Engineering and thorough-ly tested along with prototype anten-nas from Vertex/LSI and Mitsubishi at theALMA Antenna Test Facility located at the Very Large Array site in Socorro, NewMexico.

For more information on the ALMA proj-ect, see http://www.eso.org/projects/alma/

(Based on ESO Press Release 31/05)

Above: Signing event, from left to right: Mr. Thomas Zimmerer (Head Antennas and Mecha-tronic, MT Aerospace AG), Mr. Hans Steininger (ChiefFinancial Officer, MT Aerospace AG), Mr. PatrickMauté (General Manager, BU Observatorion and Sci-ence of Alcatel Alenia Space France), Dr. CatherineCesarsky (ESO Director General), Mr. Giampietro Mar-chiori (Managing Director, European Industrial Engi-neering s.r.l.), Mr. Vincenzo Giorgio (Director of Scien-tific Italian Programmes, Alcatel Alenia Space Italy),and Dr. Ian Corbett (ESO Deputy Director General).

Left: The Alcatel AleniaSpace/European In-dustrial Engineering Pro-totype Antenna.

Pho

tos:

H.Z

odet

,ES

O(2

)



Inauguration of the APEX Telescope

Gonzalo Argandoña, Felix Mirabel (ESO)

The serene Andean village of San Pedrode Atacama, in northern Chile, was the epicentre of the two-day official inau-guration of APEX, the 12-metre telescopeworking at the Llano of Chajnantor.

On September 25 and 26, representativesof the three organisations running thisantenna hosted a lively scientific celebra-tion. Dr. Catherine Cesarsky (ESO’s Di-rector General), Prof. Karl Menten (Direc-tor of the Max-Planck-Institute for RadioAstronomy and Chairman of the APEXBoard) and Prof. Roy Booth (Director ofthe Onsala Space Observatory) pre-sented the key features of this new astro-nomical facility that will provide privilegedaccess to the “cold universe”.

The intendente Jorge Molina, represen-tative in Region II of the President of ChileRicardo Lagos, remarked that ParanalObservatory, APEX and in the near futureALMA have turned this part of Chile into an active astronomical centre, with signifi-cant benefits to the local economy andthe education of the people living there.

On behalf of local communities, the Mayorof San Pedro de Atacama, Sandra Berna,celebrated the active cultural exchangeand dialogue between members of Euro-pean astronomical organisations and the inhabitants of San Pedro, most ofthem belonging to the ancient Lican Antaiculture.

Ambassadors in Chile of some of ESO’smember states, the Executive Director ofthe Chilean Science Agency (CONICYT),the Presidents of the Communities ofSequitor and Toconao, as well as repre-sentatives of the Ministry of Foreign Af-fairs and Universities in Chile also attend-ed the ceremony.

On the second day of the programme, thegroup visited the APEX base camp inSequitor, near San Pedro, from where theantenna is operated through a micro-wave link to Chajnantor. Visitors had a guided tour of Sequitor facilities, includ-ing the main control room. Here, theycould have a glimpse of some of the sci-entific results already obtained with APEX.

The telescope, designed to work at sub-millimetre wavelengths, in the 0.2- to1.5-mm range, successfully passed itsScience Verification phase in July, andsince then has been performing regularscience observations.

Above: After months of hard work set-ting up the telescope, the members of the three organisations behind APEXcelebrated at San Pedro de Atacama the beginning of regular science obser-vations. They were joined by repre-sentatives of the Chilean government,universities and local communities dur-ing the two-day event.

Among other astronomical targets, APEXwill be used for comprehensive surveys of the Galactic Plane, which will locate thesources that ALMA will study in minutedetail. In that sense, it is considered as a real pathfinder that will prepare the wayfor ALMA in the years to come.

Left: APEX is the largest sub-millimetrefacility in the southern hemisphere.The surface of the antenna has beenadjusted to an accuracy of 17 microns,less than one fifth of the thickness of a human hair, all across the surface andat all times and positions.

Pho

tos:

G.A

rgan

doña

,ES

O(to

p);A

.A.L

undg

ren,

ES

O(b

otto

m)



Towards an Automatic Reduction of FORS2-MXU Spectroscopy

MXU mode, which prompted us to con-duct a feasibility study of automatic spec-tral extraction with the aXe software.

The aXe concept

In slitless spectroscopy there are no aper-tures such as slitlets, which allow only asmall region of the sky to enter the spec-trograph. An unambiguous conversion be-tween pixel coordinates and wavelength is therefore impossible. All spectral objectextractions must be based on local solu-tions to the trace description and disper-sion solution, which are globally describedin the configuration file. To evaluate the lo-cal solutions, an object coordinate (storedin the Input Object List) must be provid-ed for every object whose spectrum willbe extracted with aXe. Taking into ac-count the optical distortions of the instru-ment, the object coordinates are trans-formed to the so-called reference positionin CCD coordinates, which drives the lo-cal solution. In addition to the object co-ordinates, the Input Object List also con-tains other parameters such as the objectextension. With this information it is possi-ble to adjust the spectral extraction widthfor each object individually.

Given a stable instrument configurationand operation, the aXe approach of aglobal calibration makes it possible toextract multi-object spectroscopy data inan extremely efficient and largely unsu-pervised way. The main task of the data-reduction is reduced to the creation of the Input Object List. Especially for theproduction of large science-ready data-products, to be stored in Virtual Observa-tory compliant archives, a software pack-age like aXe is of paramount importance.In the traditional data reduction, the ac-curacy of the instrumental calibration is limited to the accuracy of the set of cal-ibrations taken with the science data. In the aXe approach of a global calibra-tion, however, the accuracy is limited by the stability of the instrument and theaccuracy of the calibration data to ca-librate the telescope/instrument setup.

Observing with FORS2-MXU

FORS is the visual and near-UV FOcal-Reducer and low-dispersion Spectro-

Harald Kuntschner, Martin Kümmel, Søren Larsen, Jeremy Walsh (ST-ECF)

A brief report is presented of a feasibil-ity study for an automatic, sciencequality spectral extraction of FORS2-MXU spectra using the aXe packagedeveloped for slitless spectroscopy. Webriefly describe the aXe data reduc-tion concept, which is usually applied toHST-ACS data, and demonstrate theapplicability of this concept by reduc-ing a subset of the Great ObservatoriesOrigins Survey (GOODS) FORS2 spec-troscopic survey.

The large numbers of objects which canbe observed simultaneously with modernmulti-object spectrographs demand the availability of automatic reduction soft-ware. Such pipeline-supported data-reduction is often used in space astrono-my. One example is the purpose-builtsoftware aXe (http://www.stecf.org/soft-ware/aXe), which provides science-qual-ity spectra for the slitless modes of theACS instrument aboard HST. ACS showsa high degree of stability, which allows the production of a generic calibration(e.g., reference files for the trace, thewavelength solution, flat-field and sensitiv-ity curves) for each of the dispersiveelements, removing the need to calibrateeach individual set of observations. Thedata-reduction software is then only driv-en by the object coordinates within theField of View (FoV). For long exposures,the spectra of thousands of objects can be extracted, which is clearly beyondmanual effort. The software has been suc-cessfully used on ACS data in variousprojects such as the Hubble Ultra DeepField HRC Parallels (Walsh, Kümmel &Larsen, 2004), high redhift supernovaesearch (Riess et al. 2004) and the GrismACS Program for Extragalactic Science(GRAPES, Pirzkal et al. 2004). It was de-signed such that the instrument specificparameters are stored in external configu-ration files. The configuration files are de-fined in a very general and thus flexibleway. Therefore the software can be easilyadapted to new instruments, even multi-slit spectrographs such as FORS2 at theVLT. Currently there is no software re-duction package available for the FORS2-

graph for the Very Large Telescope (VLT).FORS2 offers the possibility to insert inthe focal plane a mask in which slits ofdifferent length, width and shape can becut with a dedicated laser-cutting ma-chine (MXU mode). The FORS Instrumen-tal Mask Simulator (FIMS) tool is used to design and define the masks. Withinthe FIMS tool the slit positions and anumber of reference stars are selected.These reference stars will be automat-ically identified on the acquisition imagetaken within the target acquisition se-quence. From these positions the transla-tion and rotation offsets are calculatedand transferred automatically to the tele-scope. Thus one ensures that the ob-jects are accurately located within the slitsof the mask. The FIMS tool is very ver-satile and thus allows designing masks indifferent ways. The masks are normallybased on pre-imaging with the FORS in-strument and slits are placed on objectsvisible on these images. This can be doneby hand or in a semi-automatic fashion.One can also create masks on the basisof catalogues only. Common to all ofthese methods is that the object positionis generally not in the centre (along thespatial direction) of the slit. Although theslit positions and dimensions of eachmask are carried through to the final sci-ence images, the object positions with-in these slits are not generally stored.Therefore, only the person who has creat-ed the mask may have this information.From the viewpoint of automatic data-reduction or the use of this data by otherastronomers than the original scienceteam, the absence of the object positionsis a major drawback.

Calibrating FORS2-MXU

As a test bed for our feasibility study wehave selected the GOODS/FORS2 spec-troscopic survey. This data set waschosen since it provides a rich and chal-lenging use of the FORS2-MXU mode.Additionally, we can compare with an ex-isting ”by-hand” data reduction (Vanzellaet al. 2005) and thus are able to evaluatethe quality of our data products. In Fig-ure 1 a typical mask design and the asso-ciated spectral exposure are shown. Inour feasibility study we derive a calibrationfor FORS2 only for the spectral mode ofthe GOODS data set.


In order to apply the aXe software con-cept to FORS2-MXU data, two majoritems need to be addressed: (a) the pro-duction of the Input Object List whichdrives the data-reduction tasks; and (b)the construction of the aXe configurationfiles which describe the imprint of thespectrograph on the incoming beam andother instrument characteristics such asthe detector flat field.

In the case of FORS2 one can establishthe overall sky-coordinate to CCD-co-ordinate transformation via the mandatorythrough-slit images and the known opticaldistortions of the instrument. In the ab-sence of individual object coordinates one can instead use the slit centres (andlengths) in the Input Object list and thusextract two-dimensional spectra from theraw images. An external source-findingalgorithm could then be used to extractone-dimensional spectra of all the objectswithin the individual slits. Of course, theaXe software can also be used in the nor-mal configuration if, as in the case of theGOODS survey, the object positions areaccurately known.

In order to derive a global calibration forFORS2 and the 300I grism used for the GOODS survey, we have obtainedthrough-slit, flat-field and wavelengthexposures with a special calibration mask in daytime. This mask features 47 slitsuniformly covering the typical area of sci-ence slits. The main calibrations neededfor aXe are the trace and wavelengthsolution as a function of the referenceposition. To derive the trace calibration, allobject traces were determined on the flat-field exposures. The traces over theFoV can be reasonably well approximatedby a field-dependent second-order poly-nomial yielding residuals of less than 0.15 pixel rms (1 pixel corresponds to0.25 arcsec on the sky). See Figure 2 foran example.

Using the trace solution we have exam-ined arc-lamp exposures obtained withour special calibration mask to derive a global wavelength solution. Here a field-dependent 4th-order polynomial fit isneeded to describe individual wavelengthsolutions with a typical rms of 0.13 Å.

As a check of our global wavelength cali-bration we compared the predicted wave-

lengths with arc-lamp observations takenthrough a science mask during daytime.Furthermore, we compared our wave-length predictions with the observationsof relatively isolated and strong nightskylines with known wavelength. The sys-tematic errors in the mean, absolutewavelength scale appear to be generallyless than 0.1 nm, and may be reducedeven further by adopting global shiftsbased on measurements of skylines. Moreimportantly, wavelength-dependent errorsalso appear to be small (< 0.1 nm), ex-cept perhaps at wavelengths shorter than600 nm where the error may reach 0.2 nm. Note that the average dispersionis 0.32 nm/pixel.

The quality of our trace and wavelengthpredictions can be best demonstratedwhen running the aXe software in a modein which two-dimensional spectra are ex-tracted. In this mode, fully-calibrated andrectified spectra are produced for each slit on the science mask. An example ofthis is shown in Figure 3 in which we addition ally apply a simple backgroundremoval procedure.

Since aXe was initially designed for slitlessspectroscopy data, the assumption ismade that any pixel can receive a spectralelement from any wavelength over thesensitive range of the detector and spec-tral elements. Thus in order to apply apixel-flat-field correction (a P-flat in HSTterminology), the wavelength dependenceof the flat field for each pixel must be de-termined. For the ACS this was performedby fitting the wavelength dependence,pixel-by-pixel, of flat fields taken with theavailable filters (for details see Walsh &Pirzkal 2005). The flat field used by aXe isa polynomial fit to the wavelength de-pendence of these flat fields; the flat fieldis then stored as a flat field cube witheach plane holding the coefficient of thepolynomial fit per pixel. An identicalprocedure was implemented for FORS2.

Required modifications to aXe

Some additions to aXe had to be imple-mented in order to reduce FORS2-MXUdata. The additions mostly deal with the “special” needs of slit spectroscopy

Telescopes and Instrumentation Kuntschner H. et al., Towards an Automatic Reduction of FORS2-MXU Spectroscopy

Figure 1: Left: A ”through-slit” image of a typical maskfrom the GOODS/FORS2 survey. The slits appear as short black lines in this image. Right: One 20-minspectroscopic exposure showing the spectra ob-tained with the mask at left and the 300I grism insert-ed. The typical wavelength range is 600 to 1000 nmat a resolution of R = 660 and thus the signal is domi-nated by the sky background, which is clearly visible.

500 1000X CCD

1500 2000

1960

1950

1940

1930

1920

1910

1900

YC

CD

Figure 2: An example ofour trace predictions(orange dashed lines) foran individual slit in CCDcoordinates. The plussign marks the projectedslit centre and the starsymbols indicate the pro-jected slit edges.


Figure 3: Top: The fully calibrated and rectified two-dimensional spectrum for one 20-min exposure of oneslit from the GOODS/FORS2 test data set as pro-duced by aXe is shown. Night skylines largely domi-nate the signal. Bottom: The same spectral regionafter the application of a simple background removalprocedure. A clear signature of a continuum source is visible. Note that for demonstration purposes the y-axis is heavily stretched.

Figure 4 (below): A com-parison with the GOODS team ”by hand”reduction for four gal-axies spanning a rangein red-shift from 0.24 to 3.70. The black solidline represents theGOODS team spectrum,the red line the aXe re-duction.

700 800 900 1000Wavelength [nm]

0102030

Y[p

ixel

]

700 800 900 1000Wavelength [nm]

0102030

Y[p

ixel

]

and ground-based astronomy, but do nottouch aXe core parts. (a) In contrast toACS slitless spectroscopy the finite lengthof slits limits the useful area for back-ground determination. Furthermore, thebackground estimator in aXe had to in-clude cosmic-ray detection and rejection,which is done prior to running aXe onACS data. (b) The production of a science-ready two-dimensional spectrum for each slit was introduced. This allows theuser to check results and fine-tuneparameters. These intermediate productscan even be used as starting points fordata reduction with other software pack-ages such as IRAF (see also Figure 3). (c) The possibility to extract spectra from significantly bent traces such as inFORS2-MXU also had to be added.

Comparison with GOODS “by hand” extraction

In order to verify the quality of our aXedata reduction, we compared our one-dimensional spectra with the GOODSteam “by-hand” extraction1. The testdata set consisted of 12 individual expo-sures (total = 14 400 s) taken with onemask. Each of the science exposures wastreated separately, and by driving the aXe software with accurate object coordi-nates, one-dimensional spectra for allobjects on the mask were produced. Fi-nally the one-dimensional spectra of theindividual objects were combined using a sigma-clipping algorithm.

Figure 4 compares a subset of our spec-tra with the ones extracted by the GOODSteam. The overall agreement between the aXe and GOODS team data reductionis good. We recover all important emis-sion and absorption lines, which are usedto determine the redshift of the objects.However, there are sometimes significantdifferences in the line strength of emis-sion lines. We note that in the aXe re-duction the individual one-dimensionalspectra were median scaled before com-bination, while the GOODS team com-bines the spectra with a weighting ac-cording to spectral quality and seeing.

1 We are grateful to the GOODS team for early accessto the spectra.

The different scaling methods and cos-mic-ray-rejection algorithms may accountfor some of the differences.

Concluding remarks

In this feasibility study we demonstratethat the current implementation of aXecan successfully extract one-dimensionalspectra from FORS2-MXU data of theGOODS survey. Although the reductionprocess is largely unsupervised we canproduce spectra of similar quality to the “by hand” reduction of the GOODSteam. There are significant tasks to becarried out before the software reachesthe maturity of a common-user reduc-tion package, for example, trending of thetrace and stability of the wavelength solu-tion for various grisms. Furthermore there

are a number of reduction steps which re-quire improvements such as cataloguegeneration and determination of referenceposition for each FORS2-MXU spectrum.We encourage astronomers who are in-terested in the application of this software to FORS2 to contact us (email: [email protected]) and let us know what demandstheir data would require of our software.

References

Pirzkal, N., Xu, C., Malhotra, S. et al. 2004, ApJS154, 501

Riess, A., Strolger, L. G., Tonry J. et al. 2004, ApJ607, 665

Vanzella et al. 2005, A&A 434, 53Walsh, J., Kümmel, M., Larsen, S. 2004, ST-ECF

Newsletter 36, 8 Walsh, J., Pirzkal, N. 2005, ACS instrument science

report 05-02

640620600580560540Wavelength [nm]

Nor

mal

ised

Flux

–1

0

1

2

3

4

5GDS J033226.27-275245.7 (z = 3.70)

Lyα

GDS J033232.13-275105.5 (z = 0.68)

700680660640620600Wavelength [nm]

Nor

mal

ised

Flux

0

1

2

3 [OII]

Wavelength [nm]960940920900880860840

Nor

mal

ised

Flux

–2

0

2

4

6

8

10

12

[OII]

GDS J033227.08-275401.3 (z = 1.39)

Nor

mal

ised

Flux

Wavelength [nm]

Hα

GDS J033238.82-274956.3 (z = 0.24)

820 840 860 880800780

0

1

2

3



The Virtual Observatory in Europe and at ESO

Paolo Padovani, Peter Quinn (ESO)

Virtual Observatories are moving from a“research and development” to an“operational” phase. We report on theEURO-VO, a project to build an opera-tional Virtual Observatory in Europe, and on the new Virtual Observatory Sys-tems department at ESO.

The Virtual Observatory (VO) is an innova-tive, evolving system, which will allowusers to interrogate multiple data centresand services in a seamless and transpar-ent way, to best utilise astronomical data.The main goal of the VO is to enable new science by making the huge amountof data currently available easily acces-sible to astronomers. The VO initiative is aglobal collaboration of the world’s astro-nomical communities under the auspicesof the recently formed International VirtualObservatory Alliance (IVOA; http://ivoa.net;see Figure 1).

The Astrophysical Virtual Observatory

The status of the VO in Europe is verygood. In addition to seven current nationalVO projects, the European funded collab-orative Astrophysical Virtual Observatory(AVO) project had the task of creating thefoundations of a regional-scale infrastruc-ture by conducting a research and dem-onstration programme on the VO scientificrequirements and necessary technolo-gies. The AVO had been jointly funded bythe European Commission (under the Fifth Framework Programme [FP5]) withsix European organisations (ESO, theEuropean Space Agency [ESA], AstroGrid,the Centre de Données astronomique de Strasbourg [CDS], TERAPIX, andJodrell Bank) participating in a three-year,Phase-A work programme.

The AVO project was driven by a strategyof regular scientific demonstrations of VO technology and is now formally con-cluded. AVO’s main achievements can be thus summarised:

Three science demonstrations

These were held on an annual basis, incoordination with the IVOA, for the AVOScience Working Group (SWG), es-tablished to provide scientific advice tothe project. Three very successful de-monstrations were held in January 2003(Jodrell Bank), 2004 (ESO, Garching), and 2005 (ESAC, Madrid).

First VO paper

We reported last year (Padovani et al.2004 b) on AVO’s second demonstration,held on January 27–28, 2004 at ESO,which led to the discovery of 31 new opti-cally faint, obscured quasar candidates(the so-called QSO 2) in the two Great Ob-servatories Origins Deep Survey (GOODS)fields. These results, in turn, led to thepublication of the first refereed astronomi-cal paper enabled via end-to-end use of VO tools and systems (Padovani et al.2004 a). The paper was also publicised by an ESA/ESO press release (see http://www.euro-vo.org/pub/articles/AVO1stSciencePressRelease.html ).

VO tools

For the purpose of the demonstrationsprogressively more complex AVO de-monstrators have been constructed. Thecurrent one is an evolution of Aladin, de-veloped at CDS, and has become a set ofvarious software components, providedby AVO and international partners, whichallows relatively easy access to remotedata sets (images and spectra), manipula-

tion of image and catalogue data, andremote calculations in a fashion similar toremote computing. The AVO prototype isa VO tool which can be used now for the day-to-day work of astronomers andcan be downloaded from the AVO Website as a Java application. We note thatthis is by definition a prototype and willnot be maintained on the long term. Mostof the functionalities developed for theAVO demonstrations are now available inthe public version of Aladin, and the in-clusion of the remaining ones is being as-sessed.

Science Reference Mission

The Science Reference Mission is a de-finition of the key scientific results that thefull-fledged VO in Europe should achievewhen fully implemented. It consists of anumber of science cases, with related re-quirements, against which the success of the operational VO in Europe will bemeasured. It was put together by the AVOSWG.

The AVO was also a founding member ofthe IVOA, and has provided fundamen-tal input for the development of VO stan-dards and their usage.

The AVO project is now formally conclud-ed. Links to various documents and to the software download page can befound at http://www.euro-vo.org/twiki /bin /view/Avo/.

Figure 1: The Internation-al Virtual ObservatoryAlliance, the Member Or-ganisations.


The EURO-VO

The EURO-VO work programme is thelogical next step from AVO as a Phase-Bdeployment of an operational VO in Eu-rope. Building on the development experi-ence gained within the AVO Project, incoordination with the European astronom-ical infrastructural networks OPTICONand RADIONET, and through membershipand support of the IVOA, EURO-VO willseek to obtain the following objectives:(1) technology take-up and full VO compli-ant data and resource provision by astro-nomical data centres in Europe; (2) sup-port to the scientific community to utilisethe new VO infrastructure through dis-semination, workshops, project support,and VO facility-wide resources and ser-vices; (3) building of an operational VO in-frastructure in response to new scientificchallenges via development and refine-ment of VO components, assessment ofnew technologies, design of new com-ponents and their implementation. EURO-VO is open to all European astronomicaldata centres. Initial partners include ESO, the European Space Agency, andsix national funding agencies, with theirrespective VO nodes: Istituto Nazionale diAstrofisica (INAF, Italy), Institut Nationaldes Sciences de l’Univers (INSU, France),Instituto Nacional de Técnica Aeroespa-cial (INTA, Spain), Nederlandse Onder-zoekschool voor Astronomie (NOVA, theNetherlands), Particle Physics and Astron-omy Research Council (PPARC, UnitedKingdom), and Rat Deutscher Sternwar-ten (RDS, Germany). The total plannedEURO-VO resources sum up to approxi-mately 60 persons per year over threeyears, i.e., about three times those of theAVO.

EURO-VO will seek to obtain its objec-tives by establishing three new interlinkedstructures: 1. the EURO-VO Data CentreAlliance (DCA), an alliance of Europeandata centres who will populate the EURO-VO with data, provide the physical stor-age and computational fabric and who willpublish data, metadata and services to the EURO-VO using VO technologies; 2. the EURO-VO Facility Centre (VOFC), an organisation that provides the EURO-VO with a centralised registry for re-sources, standards and certification mech-anisms as well as community support for VO technology take-up and dissemi-

nation and scientific programme supportusing VO technologies and resources; 3.the EURO-VO Technology Centre (VOTC),a distributed organisation that coordinatesa set of research and development proj-ects on the advancement of VO technolo-gy, systems and tools in response toscientific and community requirements.

The DCA will be a persistent alliance ofdata centre communities represented at anational level. Through membership in the DCA, a nation’s community of datacurators and data service providers will berepresented in a forum that will facilitatethe take-up of VO standards, share bestpractice for data providers, consolidateoperational requirements for VO-enabledtools and systems and enable the iden-tification and promotion of scientific re-quirements from programmes of strategicnational interest that require VO tech-nologies and services. Funds for the DCAhave been requested in an FP6 proposalsubmitted in September 2005.

The VOFC will provide a “public face” tothe EURO-VO. Through outreach, supportof VO-enabled science projects in thecommunity, workshops and schools, theVOFC will represent a central supportstructure to facilitate the broad take-up ofVO tools by the community. The VOFC willalso support the EURO-VO Science Ad-visory Committee (SAC) to ensure appro-priate and effective scientific guidancefrom the community of leading researchersoutside the mainstream VO projects. TheSAC will provide an up-to-date stream of high-level science requirements to theEURO-VO. The VOFC will further providecentral services to the DCA for resourceregistry, metadata standards and EURO-VO access. Funding for the VOFC has yetto be fully defined but will come initiallyfrom ESO and ESA with activities rampingup in 2006. The first VOFC activity wasthe organisation of the EURO-VO work-shop at ESO Headquarters in Garchingfrom June 27 to July 1, 2005 (Padovani &Dolensky 2005).

The VOTC will consist of a series of coor-dinated technology research and develop-ment projects conducted in a distributedmanner across the member organisations.The first project under the VOTC is theVO-TECH project, funded through the ECFP6 Proposal and contributions from the

Universities of Edinburgh, Leicester, andCambridge in the United Kingdom, ESO, CNRS and Université Louis Pasteur(France), and INAF (Italy). Additionalprojects can be brought to the VOTC viaother member organisations; one suchexample is ESA-VO. The VOTC provides a mechanism to coordinate and sharetechnological developments, a channel forDCA and VOFC requirements to be ad-dressed and for technological develop-ments to be distributed to the communityof data centres and individual scientists ina coordinated and effective manner.

The EURO-VO project will be proactive inreaching out to European astronomers.As a first step, the EURO-VO has startedmaking regular appearances at Joint Eu-ropean and National Astronomy Meetings(JENAM), as of the one in Liège in July2005.

VO and ESO

Data centres have a major role in the VO,as they are the primary source of astro-nomical data. The VO cannot (and doesnot) dictate how a data centre handles its own archive. However, a “VO-layer” isneeded to “translate” any locally definedparameter to the standard (i.e., IVOAcompliant) ones. For example, right as-cension can be defined in different waysby different data centres but the VO userneeds to know which of the many param-eters accessible through an archive in-terface is the right ascension. The longer-term vision of the VO is also to hide awayany observatory/telescope/instrumentspecific detail and work in astronomicalunits, for example, “wavelength range” andnot grism or filter name. Data providersare then advised to systematically collectmetadata (“data about data”) about the curation process, assign unique iden-tifiers, describe the general content (e.g.,physical coverage) of a collection, andprovide interface and capability parame-ters of public services. Finally, the VO will work at its best with high-level or “sci-ence-grade” data, so that the VO user isspared as much as possible any complexand time-consuming data reduction. Data centres should then make an effortto provide such data.


All these issues obviously affect ESO aswell. To address them, the Data Manage-ment and Operations Division created the Virtual Observatory Systems (VOS)Department on November 1, 2004. VOS’role is to manage ESOs involvement in VO activities and to make the ScienceArchive Facility (SAF) into a powerful sci-entific resource for the ESO community.The new department, headed by PaoloPadovani, includes, at present, the VirtualObservatory Technology (VOT) and theAdvanced Data Products (ADP) groups,led respectively by Markus Dolensky andPiero Rosati, and is made up of thirteenpeople.

The specific tasks of VOS are the following:

1. Creation of ADP, i.e., science-ready(“level-3” in ESO terms) data productsfrom the ESO archive; this will be done byusing a redesigned imaging pipeline builton top of the EIS/MVM software and also via a coordinated activity between theADP and the Data Flow System (DFS)groups to single out possible upgrades ofspecific DFS pipelines from level-2 tolevel-3. A recent ADP effort has been theprocessing via the EIS/MVM software andrelease of the ISAAC/GOODS data, which include 11600 science frames and10 600 calibration frames, photometri-cally calibrated using pre-determined zeropoints from SOFI data. These data areavailable at http://www.eso.org/science/goods/releases/20050930/ (see Figure 2);

2. Ingestion of ADP into a VO-compliantSAF; these will include not only productsgenerated by the ADP group but also by the community. For example, science-ready data products from a number of ESO projects, which are currently avail-able in a heterogeneous fashion fromcustomised web pages. In particular, asof Period 75 PIs of Large Programmesare requested to provide ESO with theirreduced data products at the time of pub-lication of their results. Moreover, PublicSurvey data products will also be ingest-ed and published in a VO-compliant way,as they will be extremely useful to VO us-ers, being highly homogeneous and wellcalibrated. These will include data fromOmegaCAM on the VLT Survey Telescope

(VST; Cappellaro 2005; Capaccioli et al.2005) and from VISTA (Emerson et al.2004);

3. Publication of ADP within the VO infra-structure; this will require a complete re-design of the SAF and of its interface. Asa first step in that direction, on the occa-sion of the opening of the ESO archive tothe world on April 4, 2005, VOS has de-ployed a redesigned archive interface. Itsaim is to facilitate data queries to archi-val users who might not be familiar withESO instruments. The new interface is ac-cessible at http://archive.eso.org/eso/eso_archive_main.html;

4. Development of VO technology, stan-dards, and tools for the ESO SAF, also viaparticipation to European VO activities, in particular through the VOTech project.

References

Capaccioli, M., Mancini, D., Sedmak, G. 2005, The Messenger 120, 10

Cappellaro, E. 2005, The Messenger 120, 13Emerson, J. P. et al. 2005, The Messenger 117, 26Padovani, P. et al. 2004a, A&A 424, 545Padovani, P. et al. 2004b, The Messenger 117, 58Padovani, P., Dolensky, M. 2005, The Messenger 121,

60

Figure 2: JHK colour composite of the ISAAC/GOODSmosaic part of the latest Advanced Data Productsgroup of the Virtual Observatory Systems (VOS) De-partment release, which includes 21 tiles in the J-, H-and Ks-bands, and five additional tiles in the J- andKs-bands (not visible in this image). Each ISAAC fieldis 2.5 arcmin across. See http://www.eso.org/science/goods/releases/20050930/

Telescopes and Instrumentation Padovani P., Quinn P., The Virtual Observatory in Europe and at ESO


Reports from Observers

GOODS’ Look at Galaxies in the Young Universe

Eros Vanzella1

Stefano Cristiani1

Mark Dickinson 2

Harald Kuntschner 7, 3

Alessandro Rettura 3, 4

Leonidas A. Moustakas 6

Mario Nonino1

Paola Popesso 3

Piero Rosati 3

Daniel Stern 6

Catherine Cesarsky 3

Henry C. Ferguson 5

Robert A. E. Fosbury 7

Mauro Giavalisco 5

Jonas Haase 7

Alvio Renzini 8

and the GOODS Team

1 INAF – Osservatorio Astronomico di Trieste, Italy

2 National Optical Astronomy Observatory,Tucson, Arizona, USA

3 ESO4 Université Paris-Sud XI, France5 Space Telescope Science Institute,

Baltimore, Maryland, USA6 Jet Propulsion Laboratory, California

Institute of Technology, Pasadena, USA7 ST-ECF8 INAF – Osservatorio Astronomico di

Padova, Italy

The Great Observatories Origin DeepSurvey is using the most advancedobserving facilities, ESO/VLT, HST,Spitzer, Chandra, XMM/Newton, etc. todiscover infant galaxies at epochs when the Universe was only 900 millionyears old. The FORS2 spectroscopicsurvey has confirmed 33 galaxies in theredshift interval 5 < z < 6.2, produc-ing one of the largest spectroscopicsamples in that redshift range to date.

The quest for primordial galactic structure

One of the major goals of contemporarycosmology is to understand the process-es leading to the formation of galaxies.According to the present standard modelthe growth of structure, from the first tinyfluctuations of matter (about one part over105 at the time of recombination, as re-corded by CMB experiments) to galaxiesand clusters of galaxies, advanced throughhierarchical mergers of dark matter con-

centrations. Eventually the gravity of thelargest aggregates grew enough to pull in and concentrate the gas needed tobuild infant galaxies. The first generationsof stars are thought to have produced the ultraviolet photons needed to reionisethe Universe ending the Dark Ages thathad followed the recombination.

Astronomers are closing the gap betweenthe CMB experiments, i.e. the epoch re-combination 380 000 years after the Big-Bang, and the observations of primordi-al galaxies, thanks to the development ofmore and more powerful instruments.Taking advantage of the imaging capabili-ties of the Advanced Camera for Surveys(ACS) onboard the Hubble Space Tele-scope and the spectroscopic power ofthe FORS2 instrument at the ESO/VLTour team has been able to identify a sub-stantial population of galaxies at redshiftsup to 6.2, when the Universe was only900 million years old. The programme hasbeen carried out in the framework of theGreat Observatories Origins Deep Survey(GOODS), which unites extremely deepobservations from NASA’s Great Observa-tories, the Spitzer Space Telescope,Hubble, and Chandra, ESA’s XMM-New-ton, and from the most powerful ground-based facilities, ESO/VLT, Keck, etc. (the GOODS project has already been de-scribed in a previous article (The Messen-ger 118, 45)).

The GOODS HST Treasury Program usesthe ACS to image the HDF-N and CDF-Sfields through four broad, non overlappingfilters: F435W (B), F606W (V ), F775W ( i )and F850LP (z). The exposure time is 3,2.5, 2.5, and 5 orbits per filter, respective-ly, reaching extended-source sensitivitieswithin 0.5–0.8 mags of the WFPC2 HDFobservations. GOODS is a deep survey,not a wide one, but it is much larger than most previous, deep HST/WFPC2 surveys, covering 320 square arcmin, 32 times the combined solid angles of theHDF-N and S, and four times larger thantheir combined flanking fields. The z-bandobservations image the optical restframelight from galaxies out to z = 1.2, with an-gular resolution superior to that fromWFPC2. The ACS BViz imaging makespossible a systematic survey of Lymanbreak galaxies at 4 < z < 6.5, reach-ing back close to the epoch of reionisa-tion.

How are the primordial galaxy candidatesselected?

High-redshift objects can be identified bysharp breaks in their spectra due tostrong absorption of the UV photons byintervening hydrogen gas. Multi-colourimaging identifies these sources as they“disappear” in images taken throughfilters sensitive to light with wavelengthsshorter than that of the break (also knownas “dropout” or “Lyman break” tech-nique). An example is shown in Figure 1,where in the upper panel two modelspectra of starburst galaxies are superim-posed on the four ACS filter transmis-sions. In this illustrative example the fluxesof the galaxies drop in the B-, and V-(and also in the i-band for the redshift 6galaxy) due to the high-redshift nature of the sources. In the bottom panel, thedirect images of Lyman break galaxies at redshifts 5.5 and 6.0 are shown: thesource at redshift 6 “disappears” in the B-, V- and i-filters. This Lyman breaktechnique for identifying distant star-form-ing galaxies has been in use for over adecade and was championed by Steideland collaborators to find z = 3 and 4 gal-axies (Steidel et al. 1999).

At redshift z < 5.5 the technique involvesthree filters: one below the Lyman limit(λrest = 912 Å), one in the Lyman forest re-gion and a third longward of the Lyman-αline (λrest = 1216 Å). For example, in this way Giavalisco et al. 2004a selectedsamples of star-forming galaxies at red-shifts between 3.5 and ~ 5.5 in theGOODS fields. At redshift > 5.5, only twofilters can be (effectively) used, since theintegrated optical depth of the Lyman-αforest is >> 1 and the break in the Spec-tral Energy Distribution (SED) is locatedbetween the i- and z-filters. The key issueis to work at a sufficiently high signal-to-noise ratio that the i-band dropouts canbe safely identified through detection in a single redder band (i.e. the z-band).This is guaranteed by the exquisiteimages obtained with the ACS onboardHST.

The expected colour ( i–z) of a star-form-ing galaxy at redshift > 5.5 increases withincreasing redshift because the break at the Lyα wavelength (1216 Å) shifts out of the i-band (see Figure 1). In this wayadopting a simple colour cut of ( i–z) > 1.3


Figure 1: An example of the Lymanbreak technique. In the upper panel twomodel spectra of starburst galaxiescalculated at redshift 5.5 and 6.0 aresuperimposed on the four ACS filtertransmissions. The bottom panel showsthe direct images of two observedLyman break galaxies at redshift 5.554and 6.095, respectively. Note how thesource at redshift 6 “disappears” in thefilters B, V and i.

4000 6000 8000 104

00.

20.

40.

60.

8Tr

ansm

issi

on

Wavelength

redshift = 5.5

redshift = 6.0

i775

V606

B435

z850

b v i z z = 5.554

it is possible to select z > 5.5 galaxies.The old stellar populations of elliptical gal-axies at redshift around 1–2 have alsovery red colours (Extremely Red Objects,or EROs) and may contaminate thesamples of colour-selected high-redshiftgalaxies. This is due to the fact that the4 000 Å break of an evolved stellar popu-lation at redshift around 1–2 moves be-yond the i-filter producing redder colours.Also Galactic cool dwarf stars can con-taminate the sample, in particular at thebright end of the magnitude distribution.Recent results from the FORS2 spec-troscopic campaign in the same field atlower redshift show that the mean ( i–z)colour for ~ 50 early-type galaxies at red-shift 1–1.3 is 0.9 ± 0.2, with a maximumof 1.25, consistent with the colour pre-dicted for a typical (L*) non-evolving ellipti-cal galaxy in that redshift interval (Vanzellaet al. 2005). The availability of a widemulti-band coverage in the GOODS field,in particular the near- and mid-infra-red observations performed with theVLT/ ISAAC and the Spitzer/IRAC facilities(from 1 to 8 microns) helps to disentan-gle between high-redshift star-forminggalaxies and lower-redshift sources on thebasis of their spectral energy distribution.

Spectroscopic confirmation and redshift measurement

Although the colour selection is very ac-curate with the exquisite data describedabove, it is still important to get a spec-trum of the candidates to measure theirredshift beyond any reasonable doubt andobtain information about their physicaland chemical properties. Reliable red-shifts provide the time coordinate neededto delineate the evolution of galaxy mass-es, morphologies, clustering, and starformation. This is precisely the goal of thespectroscopic follow-up carried out by our team with FORS2 at the ESO/VLT.I-band dropouts (and also B- and V- banddropouts) have been observed in multi-ple spectroscopic masks and co-added inorder to improve the S/N ratio. The to-tal exposure time ranges from 10 000 to ~ 80 000 seconds.

Figure 2 shows the two-dimensionalFORS2 spectra (from 6 600 Å to 10 000 Å)of 33 galaxies at redshift z > 5 (25 of themare i-band dropout sources at z > 5.3).

b v i z z = 6.095

The redshift determinations, shown at theleft-hand side of each sub-panel in thefigure, have different quality (7 are certain,13 are possible and 13 tentative) and thesources will be described in detail in aforthcoming paper (Vanzella et al., in prep-aration).

At present, the success rate of the colourselection of these very high-redshift galax-ies is 96 %. As expected and shown inFigure 3, the majority of the galaxies atredshift > 5.4 are redder than ( i–z) = 1.3.There are five sources with a ( i–z) colourbluer than 1.3. These are galaxies withredshift between 5.4 and 5.6 (the two starsymbols in Figure 3 represent twosources in the redshift interval 5.4–5.5) atthe limit of redshift selection using asimple colour cut. Part of them have beenselected as V-dropout sources. The ap-

plication of the photometric redshiftstechnique, exploiting all the photometricinformation available in the GOODS field, will improve the selection of high-zgalaxies.

Properties of the Lyman-break galaxies

The spectra are in general very blue, in-dicating the presence of young stellar populations with ages of the order of107–108 years, consistently with the UVcolour selection adopted. Figure 4 showsthe composite rest-frame UV spectrumconstructed stacking 25 emission-linespectra in the redshift range 5.1 < z < 6.2 (<z> = 5.72 ± 0.26). The Lyα emissionline, the break and the flatness of the con-tinuum redward of the Lyα are clearly visi-ble. The rest frame equivalent width of the

Reports from Observers Vanzella E. et al., GOODS’ Look at Galaxies in the Young Universe


Figure 3: Colour-magnitude diagram forthe selection of i-band dropout galax-ies. The colour cut ( i–z) = 1.3 (dashedline) outlines the region of the selection.The black dots are all sources down to z = 27.5. The open circles representobjects with redshift > 5.4 and the ar-rows indicate the 1σ lower limit in the( i–z) colour. The size of the symbolsscales with the spectroscopic redshiftvalue. Star symbols are sources with5.4 < z < 5.5. Sources with an uncer-tain spectroscopic redshift are identifiedwith a square.

Figure 4: A composite spectrum of 25 emission-line Lyman break galaxiesin the redshift interval 5.2 < z < 6.2 (the mean of the redshift distribution is<z> = 5.72 ± 0.26). The Lyα line andthe break in the continuum blueward ofthe emission are apparent.

Figure 2: A sample of two-dimensional spectra of gal-axies at z > 5 discovered during the ESO/FORS2spectroscopic survey in the GOODS-S field. The posi-tion of the Lyα line is marked with a circle and whereLyα is not present the continuum break is underlinedwith a segment. The quality of the redshift determi-

Lyα line turns out to be ~ 30 Å, compara-ble with the value measured for emission-line LBGs at redshift 3 (Shapley et al.2003). The shape of the UV continuum issuppressed shortward of the Lyα by a decrement due to intergalactic H I ab-sorption. In one galaxy the N IV]-1485 Åemission line is also detected, suggestingthat we are seeing H II regions charac-terised by very hot ionising stars (Fosburyet al. 2003). It will be interesting to com-pare the properties of our sample basedon the Lyman-break selection with thesurvey by Hu et al. 2004, which selectsLyman-α emitters on the basis of narrow-band imaging. Thanks to the large num-ber of spectroscopically confirmed galax-ies and from the ultraviolet continuum, it is possible to estimate (after correction for dust extinction) the global star-for-mation rate of the Universe at z ~ 6 (e.g.Giavalisco et al. 2004, Dickinson et al.2004) and their contribution to the reioni-sation of the Universe (e.g. Panagia et al.2005). But GOODS is much more. Fromthe VLT/ISAAC and Spitzer/IRAC ob-servations covering the electromagneticspectrum from 1.2 to 8 microns, in com-bination with the current high-redshiftsample, it will be possible for the first timeto conduct a systematic study of the stel-lar-mass content of galaxies at z ~ 6and explore even earlier formation times(e.g. Eyles et al. 2005 for two i-dropoutgalaxies).

The present sample will also allow us to study the presence of large-scalestructure and the clustering properties ofgalaxies at such primordial epochs (e.g. Lee et al. 2005) giving an indicationon the typical masses of the dark mat-ter halos in which they reside. In theseways GOODS is providing stringent testsof scenarios of galaxy formation.

References

M. Dickinson et al. 2004, ApJ 600, 99L. Eyles et al. 2005, (astro-ph/0502385)R. Fosbury et al. 2003, ApJ 596, 797M. Giavalisco et al. 2004b, ApJ 600, L 103E. Hu et al. 2004, AJ 127, 563K. Lee et al. 2005, Submitted to ApJ

(astro-ph/0508090)N. Panagia et al. 2005, ApJ 663, L1A. E. Shapley et al. 2003, AJ 588, 65C. C. Steidel et al. 1999, ApJ 519, 1E. Vanzella et al. 2005, A&A 434, 53

z = 5.087

z = 5.128

z = 5.250

z = 5.292

z = 5.362

z = 5.400 ( * )

z = 5.492

z = 5.518

z = 5.541

z = 5.554

z = 5.574

z = 5.578

z = 5.583

z = 5.600

z = 5.740

z = 5.764

z = 5.786

z = 5.787

z = 5.793

z = 5.821

z = 5.828

z = 5.830

z = 5.890

z = 5.899

z = 5.920

z = 5.939

z = 5.950

z = 5.977

z = 5.996

z = 6.000

z = 6.095

z = 6.196

z = 6.200

nation depends on the reliability of the spectral fea-tures detected (i.e. on the S/N ratio). The source at z = 5.400 (*) shows a blue continuum blueward of theLyα line, because this spectrum is a combination oftwo close sources in the slit, an i-band dropout candi-date and a lower-redshift object.

zspec = 5.5zspec = 5.8

2726252423zAB

–10

12

34

(i–

z)A

B

zspec = 6.2

1000

Composition of 25 emission line galaxies

Lyα< z> = 5.72 ± 0.26

Rel

ativ

eFl

ux0.

60.

40.

20

λ (Å )1400130012001100



The Dynamics and Evolution of Luminous GalaxyMergers: ISAAC Spectroscopy of ULIRGs

high redshift and its link to the epoch ofelliptical and QSO formation we need to first understand the details of galaxymerging and its relationship to starburstsand AGN in the nearby universe. Themost violent local mergers and the proba-ble analogus to luminous high-redshiftmergers are the ultraluminous infrared gal-axies (ULIRGs). Discovered 20 years ago with the IRAS satellite ULIRGs arenow known to be mergers of gas-richdisc galaxies. They span the full range ofmerger states beyond the first encounterto complete coalescence (see Figure 1).They are amongst the most luminous ob-jects in the local Universe, with both theirluminosities (L > 1012 LA emerging main-ly in the far-infrared) and their spacedensities similar to those of quasars. Thenear-infrared light distributions in manyULIRGs resemble those of elliptical galax-ies. They also have large molecular gasconcentrations in their central kpc regions(e.g. Downes and Solomon 1998) withdensities comparable to stellar densities inellipticals. These observational resultshave led several groups (e.g. Sanders etal. 1988) to posit that ULIRGs evolve into ellipticals through dissipative collapsetriggered by a merger. In this scenario, the mergers first go through a luminousstarburst phase, followed by a dust-en-shrouded AGN phase, and finally evolveinto optically bright, “naked” QSOs oncethey either consume or shed their shellsof gas and dust.

Numerical simulations undertaken by anumber of groups also show that theviolent merging of massive disc galaxiesproduces ULIRGs that evolve into spher-oidal remnants with properties similar to those of elliptical galaxies (e.g. Barnesand Hernquist; Naab and Burkert 2003).The simulations trace the merging pro-cess from the initial encounter to final co-alescence, when the merger remnant hassettled into dynamical equilibrium. Theypredict that soon after the first encounter,the interstellar medium of the two galax-ies is efficiently concentrated in the cen-tral few kiloparsecs on a dynamical time-scale (a few tens of millions of years) due to large gravitational torques remov-ing angular momentum from the gas.Equal-mass mergers of massive galaxiesproduce the highest central gas concen-trations. Even in these most violent merg-ers, the kinematics of the system already

reach their equilibrium values of rotationand dispersion by the time the two merg-er nuclei approach to within about onekiloparsec of each other, on a timescale of a few rotation periods (~ 108 yrs) (e.g.Bendo and Barnes 2000).

Therefore, the kinematic and structuralproperties of ULIRG mergers provide anexcellent observational means to track the merging process, and to test how andon what timescales star formation andAGN activity are triggered. In order to tracethe full potential evolutionary sequence of ULIRGs we have undertaken a spectro-scopic programme with ISAAC of a flux-limited, moderately large sample of ULIRGsand a smaller sample of Palomar-GreenQSOs. We were awarded 21 nights as aLarge Programme, and during this timeallocation we have been able to observe atotal of 38 ULIRGs and 12 QSOs in a mix-ture of visitor and service mode. The threemain goals of the programme are: (1) toinvestigate which merger progenitor con-figurations are likely to result in an ultra-luminous infrared phase; (2) to establishan evolutionary connection between ultra-luminous infrared galaxies and ellipti-cal galaxies by comparing the dynamicalproperties of late-stage ULIRGs with thoseof elliptical galaxies; and (3) to investi-gate whether there is a fundamental linkbetween ULIRGs and optically bright quasars by comparing the host and cen-tral massive black hole properties of late-merger-stage ULIRGs with those of asample of optically selected QSOs andIR-excess QSOs in the same redshift andluminosity range. The rest of this articlepresents results that address the first twoscientific goals. Our analysis of the QSOsis now in progress.

ISAAC – the ideal instrument to studymerger evolution

To derive accurate kinematic propertiesfrom stellar absorption features, we need-ed high-quality, S/N of 30–50 on thecontinuum, near-IR spectroscopy of thesample sources. For the ULIRGs weselected sources for the study largely from the 1 Jy catalogue (Kim and Sanders1998). The 1 Jy catalogue comprises acomplete flux-limited (at 60 µm) sample of118 ULIRGs compiled from a redshift sur-vey of IRAS Faint Source Catalog objects.

Linda J. Tacconi 1

Kalliopi Dasyra1

Richard Davies1

Reinhard Genzel 1

Dieter Lutz1

Andreas Burkert 4

Thorsten Naab 4

Eckhard Sturm1

Sylvain Veilleux 2

Andrew J. Baker 2, 3

David B. Sanders 5

1 Max-Planck-Institut für ExtraterrestrischePhysik, Garching, Germany

2 University of Maryland, USA3 NRAO, Charlottesville, USA4 Ludwig-Maximillians-Universität, Depart-

ment für Physik, Munich, Germany5 Institute for Astronomy, University

of Hawaii, USA

Local ultraluminous galaxy mergers pro-vide us with a quantitative observation-al means to track an important processfor galaxy formation and evolution athigh redshift. This article presents firstresults from a near-infrared spectro-scopic study of a moderately large sam-ple of local universe mergers that wehave undertaken at the VLT with ISAAC.This study is providing compelling ob-servational evidence that mergers ofnear-equal-mass, gas-rich galaxies canevolve into intermediate-mass ellipticalgalaxies after passing through an ultra-luminous infrared phase.

The role of mergers in galaxy evolution

Galaxy merging is one of the main drivingforces of galaxy evolution. In hierarchi-cal CDM models of galaxy formation andevolution, merging leads to the formationof elliptical galaxies, triggers major star-bursts, and accounts for the formation ofsupermassive black holes and quasars.Many studies have shown that the impor-tance of mergers increases with redshift.Luminous, merger-induced starbursts andAGN at high redshift provide readily ob-servable signposts for tracing out the mainepoch of elliptical galaxy and quasar for-mation.

Before we can begin to assess quantita-tively the physics of the merger process at


We have observed those sources withdeclinations < 25°, and with redshiftswhere the strong rest frame H-band stel-lar absorption lines lie in parts of the H-and K-band with high atmospheric trans-mission (z < 0.13 and z > 0.20). For thesevery dusty systems, observations in thenear-infrared are superior to those in the optical for this purpose, since opticalspectroscopy cannot penetrate to thecentres of most ULIRGs. We chose to ob-serve in the rest-frame H-band, wherethere are a host of stellar absorption fea-tures (e.g. CO ∆v = 3 bands) and gasemission lines (e.g. [Fe II]) readily available.Such observations are only possible on 8-metre-class telescopes with sensitive IRspectrometers, since the strength of theabsorption features is only a few per centof the continuum. The VLT with ISAAC in its medium-resolution mode was a win-ning combination with which to carry outthe programme.

We typically integrated for one hour onsource per slit, and observed along atleast two position angles per source. Ourgroup had previously completed a smallerpilot study of about 15 ULIRGs (Genzel et al. 2001, and Tacconi et al. 2002), alsolargely with ISAAC on the VLT, and we

have augmented the current programmesample with those sources as well. Thetotal sample thus comprises 54 ULIRGsand 12 PG QSOs, and is the most com-plete and unbiased dynamical study ofthese systems to date. The sample coversthe full range of the local ULIRG luminosi-ty function, merger state and AGN activity.Our observations of these ULIRGs haveyielded unprecedented, high-quality spec-tra, from which we have been able to de-rive stellar dynamical quantities (Figure 2).From the spectra we extract the stellarvelocity dispersion, σ, and the rotationalvelocity, Vrot, for each source. We analysethe kinematic parameters together withstructural quantities taken mostly from the1 Jy sample photometric study of Veilleuxet al (2002). Our group is also conduct-ing an HST NICMOS imaging programme(PI Sylvain Veilleux) of many of the samplegalaxies.

Dynamical evidence for major mergers

The present ULIRG sample contains 21 binary (early-stage merger) systemsthat we have observed to track the pre-coalescence merger phases, and tomeasure the mass ratios of these sys-

tems. For the binary ULIRGs we havemeasured the kinematic properties sepa-rately for each progenitor nucleus in thesystem. From the stellar dispersions androtational velocities we compute the dy-namical masses of the merging galaxies.We find that ultraluminous luminosities are mainly generated by almost equal-mass mergers; the average mass ratio ofthe binary ULIRGs is 1.5 : 1 and 68 % ofthese sources are 1 : 1 encounters. Lessfrequently, we also find 3 :1 mergers inour sample, but do not have a firm casewhere the dynamical mass ratio wouldindicate ULIRG activity being triggered ina minor merger. Mergers of mass ratio > 4 :1 typically do not drive enough gas tothe centre of the merger to generate ultra-luminous luminosities. These results are in agreement with many merger models inthe literature (e.g. Naab & Burkert 2003)that all predict that ultraluminous activityis efficiently triggered in a major merger oftwo massive, gas-rich galaxies.

ULIRGs evolve into intermediate-masselliptical galaxies

We are investigating the relationship be-tween ULIRGs and elliptical galaxies by

Figure 1: The ISAAC H-band acquisition images for aselection of the programme sources, which show the variety of merging stages of the ULIRGs and alsothe excellent conditions at the VLT while the datawere obtained. The horizontal bar in the upper leftcorner of each panel corresponds to 5 kpc at the red-shift of the source.

IRAS 13335–2612 IRAS 13451+1232 IRAS 12112+0305

IRAS 11095–0238 IRAS 01004–2247 IRAS 00397–1312


comparing the dynamical properties of thelate merging stage ULIRGs (i.e. thosecoalesced into a single system) with thoseof representative samples of local ellipti-cal galaxies. The mean stellar velocity dis-persion of the fully coalesced, singlenucleus ULIRGs is 157 (± 40) km s–1. Thismean dispersion is comparable or slightlylower than that of an L* elliptical galaxy(defined as MB ~ – 20.4 mag). The velocitydispersion distribution of the ULIRGsample, in fact, closely matches that ofcompact-core, discy-isophote ellipticalgalaxies with intermediate masses, and isvery different from the (more massive)slowly-rotating/large-core/boxy-isophoteellipticals.

As a result of the virial theorem and com-mon formation and evolution processes,dynamically hot systems (elliptical gal-axies, lenticular galaxies, and spiral galaxybulges) are all found in a well-definedplane of a space comprised of velocitydispersion (σ), effective radius (reff ), andsurface brightness with that radius (µeff ).One particularly instructive way of in-vestigating a possible evolutionary trackof ULIRGs into elliptical galaxies, there-fore, is to place both the ULIRG and ellip-tical samples on the same fundamentalplane. Placing the young, infrared lumi-nous merger systems in this space andcomparing them to much older ellipticalsis tricky, however, because of the influ-

ences of stellar evolution, extinction, andperhaps incomplete dynamical relaxation,all of which will strongly affect the centralsurface brightness and effective radii evenif they are determined in the less extin-guished near-infrared bands. To minimisethese effects, we consider only the lessevolution-sensitive, effective radius – hostvelocity dispersion (reff –σ) projection ofthe fundamental plane. We present thisprojection of the fundamental plane in Fig-ure 3 for our ULIRG sample together with samples of giant (boxy-isophotal pro-file) ellipticals, moderate-mass (discy-isophotal profile) ellipticals (both from thework of Bender et al. 1992 and Faber etal. 1997), local cluster ellipticals (fromPahre 1999), and a small sample of lumi-nous infrared galaxies (LIRGs; 1011 LA < L< 1012 LA; from James et al. 1999 andShier and Fisher 1998). The location of an elliptical galaxy along the fundamentalplane is correlated with its luminosity(mass) and its dynamical and structuralproperties. The most massive ellipticalswith boxy isophotes and large cores are found in the upper right of the diagram,while somewhat lower-luminosity, lessmassive ellipticals and lenticulars withdiscy isophotes are found in the centralpart of the plane. Figure 3 shows thatmost ULIRGs are remarkably close to oron the fundamental plane of early-typegalaxies, strongly supporting the hypothe-sis that they will ultimately evolve into

elliptical galaxies. As indicated by theirmoderate velocity dispersions and com-pact effective radii, the majority of theULIRGs populate the region of the planeoccupied by the intermediate-mass, discy-isophote elliptical galaxies. Although thelate-merger-stage ULIRGs are dynamicallyhot systems (i.e. the velocity dispersiondominates the kinematics) the ULIRGs stillshow a significant rotational component in their stellar dynamics. We find a meanrotational to dispersion velocity ratio(vrot /σ) of 0.6 for these late-stage ULIRGs.We again compare the ULIRGs to ellipti-cal galaxies (see Figure 4), and find thatULIRGs have a vrot /σ ratio comparable towhat is found for intermediate-mass ellip-tical and lenticular galaxies. We concludethat there is strong dynamical evidenceconnecting late-stage local ULIRGs overthe full range of the ULIRG luminosityfunction and intermediate-mass ellipticalgalaxies. Our previous pilot study (Genzelet al. 2001; Tacconi et al. 2002) showed a similar trend, although there we sampledonly a small range of parameter space.

Ongoing work

The detailed results from the early-merg-ing stage ULIRGs are presented in Dasyraet al. 2005, which has just been acceptedby the Astrophysical Journal. A paper in-vestigating the dynamical and black-hole

1.60 1.62 1.64 1.661.581.560.7

IRAS 23128–5919 (south)

0.9

1.1

1.3

1.5

z = 0.045Br13

[Fe II]

CO (3–6)CO (2–5) See

1.60 1.62 1.64 1.661.581.560.8

1.0

1.2

1.4

1.6IRAS 19254–7245 (south)

z = 0.062

1.60 1.621.581.561.54

IRAS 02021–2103

0.9

1.0

1.1

1.2

0.8

z = 0.12

CO (1–4)

1.66 1.68 1.70 1.721.641.620.70

0.95

1.20

1.45

1.70

z = 0.26

CO (5–8)CO (4–7)

IRAS 14070+0525

1.60

Figure 2: The normalised H-band spectra for a se-lection of the programme sources as a function of rest-wavelength. For comparison, the stellar tem-plates, convolved with the Gaussian that best re-presents the broadening function (velocity dispersion)of the galaxy are over-plotted in red.

Nor

mal

ised

Flux

Rest Wavelength

Reports from Observers Tacconi L. J. et al., The Dynamics and Evolution of Luminous Galaxy Mergers


Figure 3: The dynamical projection(effective radius – velocity dispersion,reff –σ) of the fundamental plane for our sample of ULIRGs, compared withsamples of elliptical galaxies and low-er-luminosity infrared galaxies (LIRGs).See text for references.

Figure 4: Here we show a three-dimen-sional plot (σ– reff –vrot /σ), which in-dicates the distribution of the ratio ofrotation-to-dispersion velocity for thefully coalesced ULIRGs and local ellipti-cal galaxies. The ULIRGs are shown in red, intermediate-mass (discy) ellipti-cals in green and giant (boxy) ellipticalsin gold. The elliptical galaxy propertiesare taken from Bender et al. (1992) andFaber et al. (1997).

properties of the late-merging-stage,single-nucleus systems in the sample isnow in the final stage of completion. Inthe third part of the programme, we areinvestigating the possible evolutionaryconnection between ULIRGs and localQSOs, and we are currently analysing theISAAC H-band spectroscopy of our sam-ple of 12 Palomar-Green QSOs spanningthe same redshift and luminosity range asthe ULIRG sample.

Acknowledgements

We are extremely grateful to the ESO staff for the ex-cellent support we have received during both thevisitor- and service-mode phases of this programme.

References

Barnes, J. E., Hernquist, L. 1996, ApJ 471, 115Bender, R. et al. 1992, ApJ 399, 462Bendo, G. J., Barnes, J. E. 2000, MNRAS 316, 315Dasyra, K. M. et al. 2005, ApJ, in press, astro-ph/

0510670Downes, D., Solomon, P. M. 1998, ApJ 507, 615Genzel, R. et al. 2001, ApJ 563, 527James, P. et al. 1999, MNRAS 309, 585Kim, D.-C, Sanders, D. B. 1998, ApJS 119, 41Naab, T., Burkert, A. 2003, ApJ 597, 893Pahre, M. A. 1999, ApJS 124, 127Sanders, D. et al. 1988, ApJ 328, L 35Shier, L. M., Fischer, J. 1998, ApJ 497, 163Tacconi, L. J. et al. 2002, ApJ 580, 73Veilleux, S. et al. 2002, ApJS 143, 315

2.62.42.22.01.81.61.4

–1.0

–0.5

0.0

0.5

1.0

1.5

log(σ)

log

(Ref

f(K))

Cluster ellipticalsModerate-mass, discy EsGiant, boxy EsLIRGsULIRGs

10log(reff (kpc))

log(σ(km/s)

log(

v rot/σ

)

2

–1

–1.5

–1.0

–0.5

0

0.5

2.72.5

2.32.11.9

Supernova in NGC 1559

Colour-composite image of the spiral galaxyNGC 1559 in the Reticulum constellation, ob-tained with FORS1 on the VLT. NGC 1559 islocated about 50 million light years away and isabout seven times smaller than our Milky Way.The supernova SN 2005df, discovered on thenight of August 4, 2005, is visible as the brightstar just above the galaxy. SN 2005df has beenfurther classified as a somewhat unusual typeIa supernova, caught probably 10 days beforeit reached its maximum brightness. DietrichBaade, Ferdinando Patat (ESO), Lifan Wang(Lawrence Berkeley National Laboratory, USA),and their colleagues studied its polarisationproperties and found that SN 2005df resem-bles closely SN 2001el, whose explosion wassignificantly asymmetric.

See ESO Press Photo 26/05 for more details.



Lithium Isotopic Abundances in Metal-Poor Stars:A Problem for Standard Big Bang Nucleosynthesis?

Poul E. Nissen1

Martin Asplund 2

David L. Lambert 3

Francesca Primas 4

Verne V. Smith 5

1 Department of Physics and Astronomy,Aarhus University, Denmark

2 Research School of Astronomy andAstrophysics, Mt. Stromlo Observatory,Australia

3 The W. J. McDonald Observatory, The University of Texas at Austin, USA

4 ESO5 National Optical Astronomy Observatory,

Tucson, USA

Spectra obtained with VLT/UVES sug-gest the existence of the 6Li isotope inseveral metal-poor stars at a level thatchallenges ideas about its synthesis.The 7Li abundance is, on the other hand,a factor of three lower than predicted by standard Big Bang nucleosynthesistheory. Both problems may be explainedif decaying supersymmetric particlesaffect the synthesis of light elements inthe Big Bang.

Ever since the discovery of a plateau forlithium abundances in warm, metal-poorhalo stars by Spite & Spite (1982), the Big Bang has been identified as the initialand major origin of the 7Li isotope, withthe plateau abundance being used to testtheories of Big Bang nucleosynthesis.According to the standard Big Bang mod-el, the relative abundances of the lightelements (hydrogen, deuterium, heliumand lithium) depend on only one parame-ter: the baryon-to-photon ratio η. Usingthe recent precise determination of η fromcosmic microwave background fluctua-tions, the lithium-to-hydrogen ratio ispredicted to be N7Li /NH = (4.15 ± 0.5) ×10–10 (Coc et al. 2004). This correspondsto a logarithmic abundance logε (7Li) ≡log(N7Li /NH) + 12.0 = 2.62 ± 0.05 on thetraditional astronomical scale, where the logarithmic hydrogen abundance isnormalised to 12. The predicted 7Li abun-dance is about a factor three higher than Li abundances found in very metal-poor stars on the Spite plateau. Thus, akey question has become – How doesone bridge the gap between 7Li observa-tion and prediction?

A second question has arisen from the detection of 6Li in the metal-poor star HD 84937 (Smith et al. 1993; Hobbs &Thorburn 1994; Cayrel et al. 1999) with anabundance that is orders of magnitudeshigher than predicted from standard Big Bang nucleosynthesis. How does oneexplain this unexpected high 6Li abun-dance?

In order to study these two lithium prob-lems in more detail, we have conducted asurvey of isotopic lithium abundances in asample of 24 dwarf stars ranging in metalabundance, [Fe/H] ≡ log(NFe /NH)Star –log(NFe /NH)Sun, from –1 to –3, i.e. withmetal-to-hydrogen ratios that are a factorof 10 to 1000 smaller than the ratio in theSun. Hence, the stars are likely to havebeen formed from interstellar gas relativelylittle affected by element production instars.

VLT/UVES high-resolution spectra

The isotopic shift of the 670.8 nm reso-nance line of 6Li relative to the correspon-ding 7Li line is only 0.016 nm. This is com-parable to the width of the lithium line due

to fine structure splitting and line broad-ening caused by turbulence in the stellaratmosphere. In addition, it turns out thatthe 6Li/7Li ratio does not exceed 10 %.Hence, the presence of 6Li is revealed bya slight additional asymmetry of the 670.8 nm line. Consequently, both veryhigh spectral resolution and signal-to-noise are needed to detect 6Li. With 4-m-class telescopes, only a few fairly bright (V < 9 mag) stars, such as HD 84937,could be studied, but beginning in 1999the VLT and its high-resolution spectro-graph UVES opened up the possibility to conduct a systematic search for 6Li infainter and more metal-poor stars.

The spectra used in the present studywere obtained in July 2000, February2002 and August 2004. In order to obtainthe highest possible spectral resolution (λ /∆λ = 120 000) we used an image slicer,which transforms the seeing disc of thestar to a rectangular image matching the narrow (0.3 arcsec) entrance slit of theUVES spectrograph. In addition to im-proving the efficiency of the observations,the image slicer also serves to broadenthe spectrum, and hence to minimiseproblems of flatfielding the spectra to asmooth continuum.

A detailed description of the observationsand data reduction is given in Asplund et al. (2005). The spectra cover the spec-tral region 600–820 nm, except for theAugust 2004 observations, when the verymetal-poor star LP815-43 was re-ob-served with a different UVES setting cor-responding to the 500–700 nm region in order to test a tentative detection of 6Li based on the July 2000 spectrum.

Figure 1 shows the spectra of three starsin the wavelength region around the Li 670.8 nm line. The signal-to-noise (S/N)per spectral bin (0.0027 nm) is 600, whichis typical for the sample. Note the differ-ences in metallicity between the threestars as reflected in the strength of thecalcium line at 671.8 nm.

Stellar parameters

In order to derive isotopic lithium abun-dances we must know the basic physicalparameters that characterise the atmos-phere of a given star. These parameters

G013-009, 1DFWHM6.41 km/s6.91 km/s5.91 km/s

0.75

0.80

0.85

0.90

0.95

1.00

1.05

Rel

ativ

eFl

ux

612.18 612.20 612.22 612.26612.24Wavelength [nm]

670.72 670.74 670.76 670.78 670.80 670.84670.82Wavelength [nm]

0.92

0.94

0.96

0.98

1.00

1.02

Rel

ativ

eFl

ux

6Li / 7Li0.000.050.10

7Li6Li

∆χ2

S /N = 527

0.00–0.05 0.05 0.10–0.106Li /7Li

0

5

10

15

20

FWHM6.326.256.39

G013-009, 1D

Figure 3: llustration of the method applied to deter-mine 6Li/7Li in G 013-009. See text for a descriptionof the three panels.


Fe I Fe I

BD +03°740Teff = 6270 K[Fe/H ] = –2.65

CD –48°2445Teff = 6220 K[Fe/H ] = –1.93

Ca I

CD –33°3337Teff = 5895 K[Fe/H ] = –1.31

0.60

Rel

ativ

eFl

ux

670.5 671.0 671.5 672.0Wavelength [nm]

0.65

0.70

0.75

0.80

0.90

0.95

1.00

1.05

0.85

670.0

Li I

Figure 1: Sample spectra around theLi I 670.8 nm line with CD –48°2445and CD –33°3337 shifted 0.10 and0.20, respectively. Note, that the regionbetween the Li I and the Ca I lines in the spectrum of CD –33°3337 is affect-ed by several faint lines not identifiedon the figure.

Figure 2: HR-diagram for stars with[Fe/H] < –1.7. Filled circles are starswith a ≥ 2σ detection of 6Li, and opencircles refer to stars with no clear de-tection of 6Li. Evolutionary tracks corre-sponding to a range of masses arefrom VandenBerg et al. (2000).

are: the effective temperature Teff, the sur-face gravity g, and the metallicity [Fe/H].

Precise values of Teff are particularlyimportant. Often this characteristic sur-face temperature is derived from thestellar colour, but as the observed colourmay be affected by interstellar reddeningthe value obtained can be wrong. In-stead, we have obtained Teff by comparingthe observed Hα line in the UVES spectrawith synthetic profiles calculated formodel atmospheres having a range of ef-fective temperatures. As discussed inAsplund et al. (2005), relative values of Teffare determined to a precision of about ± 30 K; the absolute Teff scale is, however,more uncertain.

The surface gravity of a star was estimat-ed via its absolute magnitude MV as de-termined from Strömgren photometry andHipparcos parallaxes. The metallicity isbased on Fe II lines. The values of theseparameters are not critical for the derivedisotopic Li abundances, but they are im-portant when discussing the interpreta-tion of our results. Thus, the MV-logTeffdiagram in Figure 2 shows that our mostmetal-poor stars (except HD 19445) lieclose to the turnoff region of halo stars.From a comparison with evolutionarytracks from VandenBerg et al. (2000) wederive ages of about 12–14 Gyr confirm-ing that the sample is indeed representa-tive of the oldest stellar population in ourgalaxy.

[Fe /H] = – 2.01[Fe/H] = – 2.31

3.90 3.85 3.80 3.75log Teff

3.707

6

5

4

3

2

MV

0.7 MA

0.8 MA

0.6 MA

HD 19445

0.9 MA

Deriving 6Li/7Li

The lithium isotopic ratio was derived bycomparing synthetic profiles of the Li I 670.8 nm line with the observed pro-file. Traditional plane-parallel model at-mospheres were applied, but it has beenchecked that 3D hydrodynamical modelsgive similar or even higher 6Li/7Li ratiosthan 1D models.

The method is illustrated in Figure 3. First,the width of spectral lines due to macro-turbulence in the stellar atmosphere andinstrumental broadening is determined as shown in the upper panel for the Ca I

612.2 nm line. Several such lines wereapplied to determine the macroturbulencevelocity and a minor variation of the in-strumental broadening with wavelength asdetermined from the width of thoriumcomparison lines was taken into account.

In the next step (middle panel) the aver-age macroturbulence velocity plus instru-mental broadening are applied in a syn-thesis of the observed Li I 670.8 nm line.The comparison between the theoreticaland observed profiles is quantified by calculating the chi-square functionχ 2 ≡ Σ(Oi – Si )

2/σ 2, where Oi and Sidenote the observed and synthetic flux atwavelength point i, respectively, and σ = (S/N) –1 is estimated in three nearbycontinuum windows. For each 6Li/7Li, thetotal Li abundance, the wavelength zero-point of the observed spectrum and thecontinuum level are allowed to vary in orderto optimise the fit and thus minimise χ 2.


The most probable value for 6Li/7Li corre-sponds to the minimum of χ 2.

In the case of G 013-009 the derivedvalue of 6Li/7Li is close to 0.05 as seenfrom the lower panel of Figure 3. Thisvalue is quite insensitive to the estimatederror in the macroturbulence velocity. The 1σ, 2σ and 3σ confidence limits ofthe determination correspond to ∆χ 2 =χ 2 – χ 2

min = 1, 4 and 9, respectively. Asseen, 6Li is detected in G 013-009 with aconfidence limit of about 2.5σ, i.e. with a probability of 99 %.

The derived 6Li/7Li ratios are plotted as afunction of [Fe/H] in Figure 4. As seen,there are nine stars with a ≥ 2σ detectionof 6Li. Many of the other stars may alsohave a small amount of 6Li; the average of6Li/7Li for the whole sample is close to2 %. In this connection, it is natural to askif there could be a systematic error in the 6Li/7Li determinations of this smallamount. If so, there would be very fewsignificant detections of 6Li. The fact thatthe main-sequence star HD 19445 has a6Li/7Li ratio close to zero (see Figure 4) is,however, an argument against this pos-sibility. As shown in Figure 2, HD 19445 isan unevolved main-sequence star. With its mass of about 0.65 solar masses, ithas a rather deep upper convection zone,and according to models of stellar struc-ture, 6Li cannot survive proton destruction in such a star. Hence the star can beconsidered as a reliable check of the zeropoint of the 6Li determinations.

Except for HD 19445, the stars in Figure 4are all turnoff stars or slightly evolved be-yond the turnoff. Particularly interesting isthe very metal-poor star LP 815-43 (Teff =6 400 K, logg = 4.2 and [Fe /H] = – 2.7).From the July 2000 spectrum, we ob-tained 6Li/7Li = 0.078 ± 0.033. Given therather large error bar, it was decided tore-observe the star in August 2004 with adifferent setting of UVES to get an inde-pendent determination of 6Li/7Li. The newresult is 6Li/7Li = 0.046 ± 0.022. Withinthe error bars the two results agree fairlywell, and the weighted average value is6Li/7Li = 0.056 ± 0.018, i.e. formally a 3σdetection of 6Li.

The two lithium problems

In addition to 6Li/7Li, we derive precisevalues for the total lithium abundancefrom the fits to the 670.8 nm line. Further-more, the exceptionally high quality of ourUVES spectra allows Li abundances to be derived from the subordinate Li I line at610.36 nm. Figure 5 shows this line in HD 19445 together with synthetic spectracalculated for various Li abundances. The best fit is obtained for logε(Li) = 2.19± 0.05, close to the value obtained from

the 670.8 nm line. The subordinate line isdetected in 22 of the 24 stars, and themean difference of Li abundances derivedfrom the subordinate line and the reso-nance line is + 0.05 ± 0.05 dex. This is anencouraging confirmation of the reliabilityof our Li abundance scale.

In Figure 6, 6Li and 7Li abundances areplotted as a function of [Fe/H]. Except forone star, HD 106038 (known to havepeculiar high s-process abundances), thescatter in 7Li is remarkably small, i.e.

Figure 4: The derived6Li / 7Li ratio as a functionof [Fe/H]. Stars consid-ered to have a significantdetection (≥ 2σ ) of 6Liare shown with filled cir-cles.

Figure 5: Observed (red)and theoretical (blue) Li I 610.4 nm profiles forHD 19445. The foursynthetic profiles arecomputed with no Li and logε(Li) = 2.0, 2.2 and 2.4, respectively.

HD 19445

[Fe/H]–2.8 –2.6 –2.4 –2.2 –2.0 –1.8 –1.6 –1.4 –1.2 –1.0

–0.06

–0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

–0.04

6Li

/7Li

LP815-43

–3.0

Reports from Observers Nissen P. E. et al., Lithium Isotopic Abundances in Metal-Poor Stars

HD 19445

1.02

1.00

0.98

0.96

0.94

Rel

ativ

eFl

ux

1.00

0.50

0.00

– 0.50

–1.00

resi

dua

l[%

]

610.25 610.30 610.35 610.40Wavelength [nm]

610.45


~ 0.03 dex at a given metallicity. The 7Liabundance appears to increase slightlywith increasing metallicity, which may bedue to Galactic cosmic-ray production of7Li. The trend implies an apparent “pri-mordial” 7Li abundance of about 2.1 dex,i.e. a factor of three below the value of2.6 dex predicted from standard Big Bangnucleosynthesis and the WMAP-basedvalue of the baryon-to-photon ratio. Peo-ple studying stellar structure and evolu-tion have tried to explain this difference asdue to depletion of 7Li by processes thatmix the gas in the stellar atmosphere withhotter layers, where Li is destroyed byreactions with fast protons. Turbulent dif-fusion seems to be the most promisingmechanism; for the right choice of freeparameters it can provide both a factor ofthree depletion and retain a very smalldispersion of 7Li among the plateau stars(Richard et al. 2005).

The turbulent diffusion models, which de-plete 7Li with a factor of three, predict,however, a depletion of 6Li by more thana factor of 30. This would mean that theobserved 6Li values in Figure 6 should becorrected upwards by at least 1.5 dex to correspond to the original 6Li abun-dance at the time of star formation. Suchhigh 6Li abundances seem impossible tobe explained by production of 6Li throughGalactic cosmic-ray processes involving α + α fusion and spallation of CNO nuclei.The production in a realistic model byRamaty et al. (2000), involving cosmicrays accelerated out of supernova ejecta-enriched superbubbles, is shown as adashed line in Figure 6; as seen the modelhas already difficulties in explaining theuncorrected 6Li abundances in the mostmetal-poor stars. More speculative mod-els of pregalactic synthesis of 6Li havebeen suggested. Thus, Suzuki and Inoue(2002) suggest that α-particles are accel-erated by gravitational shocks induced byinfalling matter during hierarchical struc-ture formation, and Rollinde et al. (2005)suggest an early burst of cosmologicalcosmic rays perhaps caused by popula-tion III stars. The energetics of such ascenario were, however, not considered.

Big Bang nucleosynthesis of lithium

The most interesting feature of Figure 6 is perhaps that the observed 6Li abun-dances appear to form a plateau in con-trast to the rise of 6Li predicted by Gal-actic cosmic-ray processes. A plateau in6Li could be an indication that 6Li has a primordial Big Bang origin like 7Li. Stan-dard Big Bang nucleosynthesis predicts,however, 6Li/7Li ~ 10–5, i.e. orders ofmagnitudes below the 6Li observations.

Extensions of standard particle physics to supersymmetry predict the existence of various exotic particles, including thegravitino. The decay of such relic particlesduring the era of Big Bang nucleosyn-thesis can alter the resulting light elementabundances, provided the masses andlifetimes of these putative particles areright. As shown by Jedamzik (2004), theinjection of energetic nucleons throughhadronic decay about 103 s after Big Bangcan lead to substantial 6Li productionwithout spoiling the agreement with ob-served values of the primordial abun-dance of D and He. Furthermore, simulta-neous destruction of 7Li with a factor 2 to 3 is possible. This would explain both

Figure 6: Observed logarithmic abundances of 6Li(red circles) and 7Li (blue crosses) as a function ofmetal abundance. Stars with a ≥ 2σ detection of 6Liare shown with filled circles; 2σ upper limits areshown for the other stars. The predicted evolution of6Li due to Galactic cosmic-ray processes (dashedline) is taken from Ramaty et al. (2000), and the pre-dicted 7Li abundance due to standard Big Bang nucleosynthesis is shown as a green horizontal line.

7Li (WMAP + SBBN)

7Li

6Li

HD 106038

–3.0 –2.8 –2.6 –2.4 –2.2 –2.0 –1.8 –1.6 –1.4 –1.2 –1.0[Fe/H ]

–0.5

0.0

0.5

1.0

2.0

2.5

3.0

3.5

1.5

log

ε(Li

)

the observed 6Li plateau and the low 7Liabundances in metal-poor stars. Thus,both of the Li problems can conceivablybe solved at the same time.

While this idea is attractive, it rests on asyet unproven physics. Furthermore, itshould be noted that the 6Li results wehave obtained are at the borderline ofbeing clear detections. Hence, more workis needed in order to obtain additionaldata especially for the most metal-poorstars and also to verify the zero-point ofthe 6Li/7Li determinations.

References

Asplund, M. et al. 2005, astro-ph/0510636Cayrel, R. et al. 1999, A&A, 343, 923Coc, A. et al. 2004, ApJ 600, 544Hobbs, L. M. and Thorburn, J. A. 1994, ApJ 428,

L 25Jedamzik, K. 2004, Phys. Rev. D 70, 063524Ramaty, R., et al. 2000, ApJ 534, 747Richard, O., Michaud, G., and Richer, J. 2005,

ApJ 619, 538Rollinde, E., Vangioni, E., and Olive, K. 2005,

ApJ 627, 666 Smith, V. V., Lambert, D. L., and Nissen, P. E. 1993,

ApJ 408, 262Spite, F. and Spite, M. 1982, A&A 115, 357Suzuki, T. K. and Inoue, S. 2002, ApJ 573, 168VandenBerg, D. A. et al. 2000, ApJ 532, 430



The VLT-FLAMES Survey of Massive Stars

Christopher Evans1, 2

Stephen Smartt 3

Daniel Lennon 2

Philip Dufton 3

Ian Hunter 3

Rohied Mokiem4

Alex de Koter 4

Mike Irwin 5

1 UK Astronomy Technology Centre,Edinburgh, United Kingdom

2 Isaac Newton Group of Telescopes, La Palma, Spain

3 Queen’s University of Belfast, Northern Ireland

4 University of Amsterdam, the Netherlands

5 Institute of Astronomy, Cambridge,United Kingdom

We have observed an unprecedentedsample of 800 massive stars in opencluster fields in the Magellanic Cloudsand Milky Way, primarily with the multi-fibre FLAMES instrument. The sur-vey addresses the role of environment,via stellar rotation and mass-loss, on the evolution of the most massive stars,which are the dominating influence onthe evolution of young, star-forming gal-axies.

The Large and Small Magellanic Clouds(LMC and SMC) are our nearest cos-mic neighbours to the Milky Way. Detailedstudies of stars and gas regions in theClouds over the past 50 years have re-vealed that conditions in them are verydifferent to those seen in our galaxy. Com-pared to the Sun, the Clouds are found to be metal-poor (“metals” meaning ele-ments heavier than helium). The metalcontent, or metallicity, in the LMC is 40 %of that found in the Sun, and drops to20 % in the SMC. These different abun-dances mean that the Clouds are anexcellent laboratory to test our under-standing of the role of metals in star for-mation and stellar evolution.

Interest in massive stars in the MagellanicClouds has been particularly strong inrecent years. The evolution of O and earlyB-type stars is dominated by the ef-fects of mass-loss from their strong stellarwinds, with the most massive O starsalready losing a significant amount of theirinitial mass over their core hydrogen-burning lifetimes. Stellar winds are accel-erated by momentum transfer fromphotons in the radiation field to metal ions(such as carbon, nitrogen, and iron) in the outer atmosphere of the star. It followsthat the intensity of these winds, andtherefore the mass lost by a star over itslifetime, are expected to be dependent onthe metallicity of the region in which thestar formed (Kudritzki et al., 1987). Obser-vational evidence of this has been rela-tively scarce, limited to small samples ofO stars (Puls et al. 1996) and luminous B-type supergiants (Evans et al. 2004;Crowther et al. 2006).

Another factor that strongly affects theevolution of a star is its rate of rota-tion. Stellar evolutionary models that in-clude the effects of rotation predictenhanced amounts of helium and nitrogenat the surface of the atmosphere. Theimportance of this process is thought to depend on the initial metal abundance

(Maeder and Meynet, 2001). There arealso suggestions that the distribution ofstellar rotation rates may depend on the metal abundance (Maeder et al. 1999).

Understanding the physical processes in massive stars has far-reaching implica-tions, from the feedback of kinetic en-ergy into the interstellar medium and met-al enrichment of their host galaxies, to issues such as the progenitors of su-pernova explosions and gamma-raybursts. The unique combination of high-resolution, multi-object spectroscopy from FLAMES provides us with an excel-lent opportunity to expand on previousstudies, which were limited to a few tensof stars by the available instrumentation.The FLAMES Survey of Massive Stars hasobserved a large sample of O and B-typestars, in a range of environments (i.e. theMilky Way, the LMC and the SMC) to fullyinvestigate the role of metallicity on stellarevolution.

Target fields

Our FLAMES fields were centred onseven stellar clusters, selected to samplea range of age and metallicity as summa-rised in Table 1. Our targets were selectedfrom images taken with the Wide-FieldImager (WFI) on the 2.2-m telescope at La Silla, most of which were from the ESOImaging Survey pre-FLAMES programme.We have observed 750 stars withFLAMES, using six of the standard wave-length settings of the Giraffe spectro-graph at high resolution (R ~ 20 000). Thisgives continuous wavelength coveragefrom 385–475 nm in the blue, with addi-tional coverage from 638–662 nm. Thered region includes the Hα Balmer line, animportant diagnostic of mass-loss fromstellar winds and invaluable for identifica-tion of Be-type stars. The brightest 50stars in the three Milky Way clusters wereobserved separately with FEROS on the2.2-m. The survey is introduced at length

Milky Way

LMC

SMC

Metallicity

Solar

0.4 * Solar

0.2 * Solar

“Young clusters”(< 5 Myrs)

NGC 6611

N11

NGC 346

“Old clusters”(10–20 Myrs)

NGC 3293 and NGC 4755

NGC 2004

NGC 330

Table 1: Summary of fields observed with VLT-FLAMES.


by Evans et al. (2005), together with adiscussion of the Milky Way data. A simi-lar paper giving a thorough overview ofthe LMC and SMC data will also be pub-lished in due course. The distribution ofspectral types in the whole survey is sum-marised in Table 2.

A V-band WFI image of the N11 region inthe LMC is shown in Figure 1. Most of the FLAMES targets are highlighted, dis-tinguishing between O and B-type stars.There is significant structure in the N11region, with many subtle gaseous arcsand filaments seen only in the near-IR. Thedense nebula to the north of centre isLucke-Hodge 10 in which, as one mightexpect, a number of O-type stars arefound.

The hottest normal stars are those classi-fied as types O2 and O3, with three ofthese discovered in Lucke-Hodge 10 byParker et al. (1992). There are still only ~ ten O2-type stars known anywhere, sothe statistical sample for attemptingstudies of these objects remains small.Star #26 in our N11 field (marked in Fig-ure 1, with the FLAMES spectrum shownin Figure 2) is particularly interesting and is classified as O2.5 III (f*). Aside fromproviding a further example of one ofthese extreme objects, its location awayfrom the centre is also of note. The gen-eral consensus is that this region is

Field

NGC 3293

NGC 4755

NGC 6611

NGC 330

NGC 346

NGC 2004

N11

Total

O

–

–

13

6

19

4 (+ 1 WR)

43

86

Early-B(B0-3)

48

54

28

98

84

101

77

490

Late-B(B5-9)

51

44

12

11

2

6

–

126

AFG

27

10

32

10

11

7

4

101

Total

126

108

85

125

116

119

124

803

Table 2: Overview of the distribution of spectral types in the FLAMES survey.

Figure 1: V-band WFI image of FLAMES targets inN11 in the LMC. O-type stars are marked as openblue circles, B-type stars by yellow circles. Star #26 isan early O-type star in the north of the field, with itsspectrum shown in Figure 2. The solid black lines aresimply the gaps between CCDs in the WFI mosaicarray.

Figure 2: Two FLAMES spectra of note. The linesidentified in N11-026, from left to right, are N IV λ4058;Si IV λλ4089-4116; N V λλ4604-4620; N III λλ4634-4640-4642. The lines identified in NGC 346-023 areFe II λλ4549-4556-4584-4629. For clarity both spec-tra have been smoothed by a 1.5 Å FWHM filter.

a two-stage starburst. Star #26 and theother O stars beyond the central region of the field are likely members of this“second generation” of star formation.

Naturally we have found a number of Be-type stars in the survey, with somedisplaying permitted Fe II emission lines.The morphology of both the Hα and Fe II profiles suggest that we are observingthe stars from a range of perspectives.Single-peaked emission is seen in some,compared to others which show twin-peaked profiles, usually interpreted as ob-serving the star ‘edge-on’ through acircumstellar disc. One of these Be-typestars in our NGC 346 field is shown inFigure 2. These high-resolution data willprovide new insights into the physicalproperties and nature of Be-type stars.

The FLAMES data are uniquely powerfulin another regard. With repeat obser-vations at each of the different wavelengthsettings, the survey took over 100 h of VLT time to complete. The survey wastherefore undertaken in service mode,entailing observations at many differentepochs in Periods 71 and 72. This meansthat we are very sensitive to the detec-tion of binaries, enabling firm lower limitsto be put on the binary fraction in ourfields, of interest in the context of star for-mation and the initial mass function –such spectroscopic monitoring has rarelybeen done before, and certainly not with such high-quality data as that fromFLAMES. In some cases the spectro-scopic data are sufficiently well-sampledenough to yield periods, and to con-strain the properties of the individual com-ponents. The FLAMES data provide us

NGC 346-023

N 11-026 O2.5 III(f*)

Be

Fe II

N IIIN IV Si IV

N V

40000.5

1.0

1.5

4100 4200 4300 4400Wavelength (Å)

4500 4600 4700

2.0


with an excellent resource to contribute tostudies considering the evolutionary ef-fects of binarity, particularly with regard tomass-transfer of processed elements.

Atmospheric analysis

The use of self-consistent methods and a uniform dataset are the key asset of theFLAMES survey. However, by their verynature O and B-type stars warrant differ-ent approaches in terms of analysis. In O stars the stellar winds are a significantfactor to be considered when attemptingto synthesise their observed spectra,requiring more sophisticated atmosphericmodels than the majority of B stars.

A total of 86 O-type stars were observedin the survey, nearly half of which werepreviously unknown. Those in the LMCand SMC fields are now being analysedwith one of the state-of-the-art modelatmosphere codes, employing a newautomated approach with genetic algo-rithms (Mokiem et al. 2005). With multi-object instrumentation yielding ever largersamples in all fields of astronomy, suchautomation is becoming increasingly rele-vant.

Table 2 reveals that the dominant compo-nent of the survey is a large number ofearly B-type stars. These span a range of luminosity classes and rotational velo-cities. The analysis of the B stars in thethree Milky Way clusters will be presentedby Dufton et al. (in preparation), in whichbasic physical parameters such as ef-fective temperatures and gravities are de-rived, enabling precise determination of the projected rotational velocities. Thenext step is to determine the velocitydistributions for the LMC and SMC starsusing similar methods, to investigatewhether any evidence for a metallicity de-pendence is seen. One by-product of the atmospheric analyses by Dufton et al.are estimates of the stellar masses of the sample. Figure 3 shows the mass dis-tribution for NGC 3293 and NGC 4755,which includes all stars in the spectralrange B0–B8.

Narrow-lined stars (i.e. those with lowprojected rotational velocities) are sim-pler to analyse than those rotating morequickly. In addition to basic properties

Figure 3: Distribution of stellar masses in NGC 3293and NGC 4755, showing all stars with spectral typesearlier than B9. The most massive objects are early B-type supergiants and giants, which have evolvedfrom hotter main-sequence stars.

Figure 4: FLAMES spectrum (black line) of the nar-row-lined star N11-072, classified as B0.2 V. A modelspectrum from the model atmosphere code TLUSTYis shown above in red, for which Teff = 28 800 K and logg = 3.75. Both spectra have been smoothedby a 1.5 Å FWHM to aid clarity, and the model hasbeen convolved by a rotational broadening function of15 km/s.

0 5 10 15 20 25

020

4060

Num

ber

ofob

ject

s

Stellar mass

N 11-072TLUSTY Model

40000.50

1.00

1.25

4100 4200 4300 4400Wavelength (Å)

4500 4600 4700

1.75

BO.2 V

FLAMES

0.75

1.50

Reports from Observers Evans C. et al., The VLT-FLAMES Survey of Massive Stars

such as temperature and gravity, chemicalabundances of metals can be foundbecause of the well-resolved, narrowlines. Our initial study of the B stars in theMagellanic Clouds has analysed 35 ofthose with narrow-lined spectra in theN11 and NGC 346 fields (Hunter et al., inpreparation). In Figure 4 we show a sam-ple FLAMES spectrum and the model that best matches the observations. Suchcomparisons provide a wealth of informa-tion regarding the evolutionary effects onstellar abundances. Furthermore, analy-sis of the least evolved stars also providesdiagnostics of the baseline, primordial

abundances in the Clouds, complement-ing contemporary studies of A-type starsand H II regions.

References

Crowther P. A. et al. 2006, A&A in press, astro-ph/0509436

Evans C. J. et al. 2004, PASP 116, 909Evans C. J. et al. 2005, A&A 437, 467Kudritzki R.-P. et al. 1987, A&A 173, 293Maeder A. et al. 1999, A&A 346, 459Maeder, A. and Meynet, G. 2001, A&A 373, 555Mokiem M. R. et al. 2005, A&A 441, 711Parker J. W. et al. 1992, AJ 103, 1205Puls J. et al. 1996, A&A 305, 171



Why are G and K Giants Radial Velocity Variables?

Michaela Petronilla Döllinger 1

Luca Pasquini 1

Artie Peter Hatzes 2

Johny Setiawan 3

Licio da Silva 4

Jose Renan de Medeiros 5

Oskar von der Lühe 6

Leo Girardi 7

Maria Pia Di Mauro 8

Achim Weiss 9

Markus Roth6

1 ESO2 Thüringer Landessternwarte Tautenburg,

Germany3 Max-Planck-Institut für Astronomie,

Heidelberg, Germany4 Observatorio Nacional, Rio de Janeiro,

Brazil5 Departamento de Fisica, Universidade

Federal do Rio Grande do Norte, Natal,Brazil

6 Kiepenheuer-Institut für Sonnenphysik,Freiburg, Germany

7 INAF – Osservatorio Astronomico di Trieste, Italy

8 INAF – Istituto di Astrofisica Spaziale e Fisica Cosmica, Roma, Italy

9 Max-Planck-Institut für Astrophysik,Garching bei München, Germany

G and K giants are low- and intermedi-ate mass stars that have evolved off themain sequence. Almost 20 years agothey were shown to be Radial Velocity(RV) variables with amplitudes of up to 300 ms–1. After several years of ob-servations with ESO’s Fibre-fed, Ex-tended Range, Echelle Spectrograph(FEROS) at La Silla and with the Tauten-burg 2-m telescope, we have found that three mechanisms (pulsations, plan-etary companions, rotational modula-tions) contribute to the RV variability ofthese stars.

In the last two decades, the dramaticincrease in RV precision from several hun-dreds to a few ms–1 has led to the dis-covery of RV variability in stars previouslythought to be constant. G-K giants areexcellent examples of this transforma-tion. These stars occupy a wide region of the cool portion of the Hertzsprung-Russel (H-R) diagram. Low- and interme-diate-mass (1–5 MA) stars that have mi-grated off the main sequence will spendsome hundred million years in this re-gion, either evolving along the Red GiantBranch (RGB), burning helium in the core(clump stars), or climbing the Asymptot-ic Giant Branch (AGB). In the past, manyG-K giants were used as RV standardstars, but over 15 years ago it was discov-ered that several of them were RV vari-able. Subsequent investigations of a fewobjects have established that they showmultiperiodic variability on large timescales, from one day to over 600 days.

Are all G-K giants RV variable? Why do these objects vary with such diverse periods?

The short-period (1–10 days) variations arecertainly due to oscillations where pres-sure is the restoring force (p-mode oscil-lations). The long-period variations can beexplained by the presence of stellar/substellar companions orbiting around thegiant star. However, this is not the onlypossible explanation since the variationsmay also result from so-called rotationalmodulation. If a large surface inhomo-geneity (for instance a starspot) passesthe line-of-sight of the observer as the star rotates, distortions of the spectralline profile may result which will be de-tected as an RV variation with the rotationperiod of the star. The fact that RV vari-ability in giants has higher amplitudes(50–500 ms–1) than that commonly seenin main-sequence stars suggests that this may be related to some specific char-acteristics of these stars. For instance,giants have lower surface gravities andmore extended atmospheres than main-sequence stars, and this may resultin pulsations with higher amplitudes.

Although K giant RV variability can becomplicated, it is possible to distinguishbetween these three mechanisms (pul-sations, rotational modulation, or compan-

ions) by analysing other stellar parame-ters. For instance, the characteristics of stellar oscillations are expected to varywith stellar gravity. The presence of large inhomogeneities is expected to pro-duce additional spectral features, such as the variation of chromospheric activityindicators in phase with the RV periodand asymmetries in the line spectral shape(bisector). RV variations due to com-panions, on the other hand, should notproduce changes in the line shape orspectral features measuring the level ofchromospheric activity.

In October 1999 we started a spectro-scopic survey of a sample of 83 G and K giants with the high-resolution (R =48 000) spectrograph FEROS at the ESO1.5-m telescope (1999–2002). Obser-vations have continued on a less regularbasis under Max-Planck-Gesellschaft(MPG) time on the 2.2-m MPG/ESO tele-scope at La Silla. The long-term FEROSRV accuracy at the 1.5-m telescope overthese years is of 23 ms–1, well within theinstrument specifications of 50 ms–1. Thelarge spectral coverage of the spectro-graph also allowed us to record chromos-pheric indicators such as the H- and K-lines of Ca II.

This survey covers a large part of the coolsection of the H-R diagram, including a broad region of the red giant branch(RGB) from the stars of luminosity class IVto II, including RGB stars, and clumpgiants. This programme found 13 newspectroscopic binaries. The majority of the other stars in the sample (63 %)showed RV variability above the measure-ment error that seemed to increase withstellar luminosity. About 20 % of the sam-ple was constant within the measurementaccuracy.

Accurate distances to stars as measuredby the HIPPARCOS satellite allowed us to derive absolute magnitudes, and in Fig-ure 1 we plot the degree of variabilityobserved (as measured from the scatterof the RV variations) versus absolute vis-ual magnitude (from Setiawan et al.2004). In this picture the stars with low-mass companions (planets or browndwarfs) are indicated with filled symbols.Confirmed binaries have been excluded.The trends of increasing RV variationswith absolute magnitude are clearly seen.


Figure 1: RV variability versus Mv forFEROS giants. The sample has beencleared of binaries. Stars which are candidates to host exoplanets areshown as filled triangles. Two starshosting brown dwarfs are shown asfilled circles. They stand out of the main“low level variation” group. A fraction of the stars shows no variability at theaccuracy of 23 ms–1. Spread andamplitude of the variations seem to in-crease for more luminous stars.

Figure 2: Bisector velocity span versusradial velocity for a K giant from our target sample. A clear correlation is present, indicating the presence oflarge surface inhomogeneities.

20

100

200

300

σ RV

(ms

–1)

40–2–4MV (mag)

Rotational modulation of stellar surfaceinhomogeneities

In order to unveil the nature of the RVvariations we measured the changes in the photospheric line shapes by meas-uring the line bisector. In addition, wemeasured chromospheric activity usingthe asymmetry of the Ca II K line core,which is an excellent chromospheric indi-cator (see e.g. Pasquini et al. 1988).

We found eight G and K giants whose RVvariations correlate with changes in theasymmetry in the spectral line profiles (linebisectors) and with changes in the chro-mospheric Ca II K line emission core. Thebisector velocity span versus radial ve-locity is shown in Figure 2 for one of theseobjects: HD 81797.

Several of these stars have large diame-ters, up to 20 milliarcseconds, whichcould be easily resolved by the ESO VeryLarge Telescope Interferometer (VLTI).They are therefore excellent candidatesfor VLTI observations, in which the ef-fect of the surface inhomogeneities canbe detected from detailed investiga-tions of the fringe contrast as a function of wavelength.

AMBER, the near-infrared/red focal instru-ment of the VLTI, can use three tele-scopes (thereby yielding information onspot geometry in addition), with base-lines providing up to 1 mas angular reso-lution, and a spectral resolution up to10 000.

Low-mass companions

Doppler shifts of spectral lines caused by low-mass companions are neitherexpected to induce any variations in thespectral line shape nor to be accompa-nied by variations in stellar activity indica-tors. By measuring the projected rota-tional velocity of the stars and estimatingtheir radius we can also derive the rota-tional period and check whether it differssubstantially from the orbital period of the companion. This will further enable usto exclude rotational modulation as apossible cause of RV variations. We haveso far identified three stars which possesscompanions (Setiawan et al. 2005).

These cool evolved stars lie in the clumpregion, where stars undergo core heliumburning after having ascended the RGB.Clump giants spend 10–20 % of theirmain-sequence lifetime in this evolutionarystage. The comparison with theoreticalevolutionary tracks allows a precise deter-mination of the stellar mass, which we de-termined to be about 1.9 MA.

This result is very interesting, because itshows that by analysing evolved stars we are able to explore planet formation ina different range of stellar masses thanwhat is sampled in radial velocity surveysaround main-sequence stars, which is lim-ited to stars less massive than ~ 1.3 MA.

Our high-quality FEROS data allow us toderive quite accurate metallicities for our sample stars. In addition to the metal-licity [Fe/H], the effective temperature and surface gravity (logg) have been de-termined. Surprisingly, we found that the three stars proposed to host planetsare not metal-rich. Studies of host stars of exoplanets around main-sequencestars show that these tend to have highermetallicites than stars that do not possessexoplanets (e.g. Santos et al. 2004). G-K giants seem to go against this trend.However, a much larger sample of plan-ets harbouring K giant stars is needed toestablish any trend with metallicity.

Reports from Observers Döllinger M. P. et al., Why are G and K Giants Radial Velocity Variables?

250200150100500–50–100–150–200Relative Radial Velocity (m/s )

Bis

ecto

rV

eloc

ityS

pan

(m/s

)

0

50

100

150

HD 81797


Figure 3: Example of long-term varia-bility for one giant observed at the TLSobservatory. The fit represents anorbital solution for an exoplanet com-panion with P = 276 d, e = 0.48 and M sin i = 8 MJupiter.

Pulsations

Asteroseismology is a powerful tool thatuses stellar oscillations to probe theinternal structure (e.g. sizes of convectivecores) of stars and thus provides tests ofstellar structure and evolution theory.Asteroseismology is expected to providefurther information about stellar ages and masses in the near future. Althoughgreat advances have been made in he-lioseismology, asteroseismology is still inits infancy and it has been applied withsuccess to some solar-type stars, whitedwarfs, and rapidly oscillating Ap stars.More recently, precise stellar radial-veloc-ity measurements with a precision ofbetter than 1 ms–1 have discovered solar-like pulsations in η Bootis (Kjeldsen et al.2003), a G0 subgiant. Current investi-gations have detected quite short periodsin K giants. There exists evidence that os-cillations as short as a few hours arepossible. Frandsen et al. (2002) detectedoscillation periods as short as 2–5 hoursin the giant ξ Hya.

We also do not know the lowest ampli-tude for oscillations in K giants. The RV precision of FEROS may have justbeen insufficient to probe the lowest RV amplitudes, because about 20 % ofour investigated sample showed no var-iations above the measurement error (23 ms–1). Since the K giant α Ari (Kim et al., submitted) has recently beenshown to be a pulsating star with a periodof 0.84 day and an amplitude of only 20 ms–1, our “constant” K giants may justbe low-amplitude variable stars. We haveindeed indications that some of our starsmay have short-period variability.

In order to be effective for asteroseismol-ogy we must detect as many pulsationmodes (periods) as possible (the so-calledoscillation spectrum). This requires ahigher radial-velocity accuracy than hasbeen obtained with FEROS as well asbetter time sampling of the observations.Oscillation modes reach down to differ-ent depths in the stellar atmosphere de-pending on the period. If we detect manyfrequencies we can sound the interior of the star and determine stellar funda-mental parameters such as its mass andevolutionary status. These can be used to directly test theoretical models.

Beyond FEROS

We have shown that by determining thelong-term and short-period RV varia-tions in a sample of evolved stars we canlearn about planet formation, surfacestructure, and pulsations in intermediate-mass stars. Our southern sample of G-Kgiants is sufficiently large to establishwhat fraction of G-K giants are short-peri-od variable stars. Those showing signifi-cant night-to-night variations will form agroup suitable for more detailed investiga-tions by multi-site campaigns with in-creased time resolution and coverage. Forsome of our purposes a substantial in-crease in accuracy with respect to whathas been obtained with FEROS at the1.5-m telescope is needed; the high-reso-lution spectrograph HARPS, the HighAccuracy Radial velocity Planet Searcher,at the 3.6-m at La Silla is the ideal instru-ment for this follow-up.

Work is also being undertaken in thenorthern hemisphere. In February 2004we started a programme to observe asample of 62 K giants from TautenburgObservatory (TLS), in Germany. Pre-cise stellar radial-velocity measurementswere made using the echelle spectro-graph of the 2-m telescope and an iodineabsorption cell. We have continued ourprogramme for these stars with observa-tions typically made every other month.After 19 months of observations we havesome preliminary results.

These show a typical RV precision of 3–5 ms–1 which is considerably betterthan our FEROS survey. So far 60 %

of the sample shows short-period (night-to-night) variations. About 15 % of thesample exhibit long-term low-amplitudevariations and several of these may be due to planetary companions. As anexample, Figure 3 shows the long-termvariability for one giant observed at the TLS observatory. The fit represents asolution with an exoplanet companionhaving P = 276 d, e = 0.48 and M sin i =8 MJupiter.

In the TLS sample only 9 % of the starsseem to be constant. We interpret this asan evidence that most G-K giants in-deed are RV variable and that the higheraccuracy of the Tautenburg survey is the reason for the difference with theFEROS statistics. 11 stars (16 %) belongto binary systems.

In spite of our progress, we would still liketo answer the important question: Whatmakes a G-K giant pulsate and Radial Ve-locity variable?

References

Frandsen, S., Carrier, F., Aerts, C. et al. 2002, A&A 394, L5

Kim, K. M., Mkrtichian, D. E., Lee, B. C., Han, I., and Hatzes, A. P., A&A, submitted

Kjeldsen, H., Bedding, T. R., Baldry, I. K. et al. 2003, AJ 126, 1483

Pasquini, L., Pallavicini, R., Pakull, M. 1988, A&A 191,253

Santos, N. C., Israelian, G., Mayor, M. 2004, A&A 415, 1153

Setiawan, J., Pasquini, L., da Silva, L., 2004, A&A, 421, 241

Setiawan, J., Rodmann, J., da Silva, L. et al. 2005, A&A 437, L 31

–200

Period = 276 dM sin i = 8 MJupiter

e = 0.48

Rad

ialV

eloc

ity(k

m/s

)

JD-2440000053300 53400 53500 53600532005310053000

–400

0

200

400


Other Astronomical News

ASTRONET: Towards a Strategic Plan for European Astronomy

stricting our future development, we mustprove able to transcend our own mentalbarriers and present a unified front to theoutside world.

Motivated by the clear and urgent need to bring European forces together for thedevelopment of astronomy at the Euro-pean level, funding agencies and ministriesfrom France, Germany, Italy, the Nether-lands, Spain, the United Kingdom, plusESA, ESO and NOTSA, decided earlyJanuary 2005 to propose a programme to coordinate a strategic planning exercisefor European Astronomy.

This so-called ASTRONET initiative wassubmitted to the European Commis-sion (EC) early March 2005 under the ERA-NET Instrument of Framework Pro-gramme Six (FP6). The proposal wasaccepted by the EC, and ASTRONETstarted September 1, 2005. It involves ahuman effort of 299 person-months, and the total budget is 3.9 M€, includingan EC contribution of 2.5 M€. The dura-tion of the programme is four years.

ASTRONET objectives

ASTRONET covers all astrophysical ob-jects from the Sun and Solar system to the global structure of the Universe, aswell as every observing approach, inspace and from the ground and from thedetection of photons at any wavelength to astroparticles and gravitational waves.It addresses the whole scientific “foodchain” from infrastructure and technologydevelopment to observation, including thevirtual observatory, modelling, and theory.

The ASTRONET project is divided intofour main activities (work packages):

1. Networking

This work package deals with exchangeof information between ASTRONET andother relevant partners in European astro-nomical research. As a prime goal ofASTRONET is to establish regular coordi-nation between programme managersthroughout all of European astronomy,two important tasks are to integrate newparticipants early in the life of the proj-ect, and to define mechanisms to main-

Anne-Marie LagrangeASTRONET Coordinator (CNRS/INSU, France)

Establishing a comprehensive long-termplan for the development of Europeanastronomy – optical and radio, groundand space – has been discussed formany years, but still waits to be done.Also, European astronomy would ben-efit much from better-coordinatedactions among countries. A consortiumof European funding agencies is nowlaunching ASTRONET, a four-year ERA-NET initiative to achieve these goals.

European astronomy has achieved manysuccesses in the second half of the twen-tieth century, especially by pooling itspreviously scattered efforts into multilater-al partnerships. The most important ofthese are ESO for ground-based opticalastronomy and ESA for space astronomy.

If we wish to be at least as successful inthe future, we need to reinforce these jointefforts and expand them to all domains of astronomy. This requires a shared Sci-ence Vision throughout Europe and anagreed roadmap for infrastructures in as-tronomy. We also need to identify thebarriers which impede collaborative proj-ects among countries and make propos-als to overcome them.

Why invent ASTRONET?

First, the “European astronomy” referredto above is essentially confined to themember states of ESO and ESA. Astron-omers in the new member states of theEuropean Union have largely been left outof this development. And they repre-sent an intellectual capital no less valuablethan that of their more fortunate col-leagues.

Second, while long-term plans are beingprepared by ESO and ESA, they do notcover the whole field by any means. Apartfrom ALMA, planning for radio astronomyis not integrated with these plans, andlittle if any coordination exists between theplans for ground- and space-based as-tronomy. Surely, if European astronomy isto overcome the funding limitations re-

tain the Europe-wide, cross-disciplinarycoordination initiated by ASTRONET on apermanent basis.

2. A Science Vision for European astronomy

This work package aims at developing aglobal census of European astronomi-cal resources and national strategies, fol-lowed by a science-oriented StrategicReview for the next 15–20 years. TheReview is to be conducted by a ScienceVision Working Group with input from,and endorsed by, a wide European astro-nomical community by March 2007.

3. A roadmap for the development ofinfrastructure for European astronomy

Based on the recommendations of the“Science Vision”, this third work packagewill produce a strategic plan for the coordinated development of space- andground-based astronomy in Europe,including the identification of key enablingtechnologies, by December 2008. It willalso identify and initiate concrete mecha-nisms to implement this roadmap.

4. Targeted coordinated actions tostrengthen astronomy and astrophysicsin Europe

This work package will identify formal bar-riers to the further development of Eu-rope-wide cooperation and initiate coordi-nated actions to strengthen astronomy in Europe, through the development ofcommon evaluation procedures and thelaunch of a specific multi-agency researchprogramme.

Who is ASTRONET?

ASTRONET has, so far, two levels ofinvolvement, participants and associates.Participants are responsible for fulfillingthe ASTRONET programme. Associatesparticipate fully in the work programme,but, contrary to participants, do not leadany task and do not manage EC funding.


As of the start of the programme, parti-cipants are: (1) Institut National des Sciences de l’Univers du CNRS(CNRS/INSU, France); (2) Bundes-ministerium für Bildung und Forschung(BMBF, Germany); (3) ESO; (4) IstitutoNazionale di Astrofisica (INAF, Italy); (5)Particle Physics and Astronomy ResearchCouncil (PPARC, United Kingdom); (6)Nordic Optical Telescope Scientific Asso-ciation (NOTSA); (7) Ministerio de Educa-ción y Ciencia (MEC, Spain); 8) Neder-landse Organisatie voor WetenschappelijkOnderzoek (NWO, the Netherlands) and(9) Projektträger DESY (PT-DESY, Ger-many). Associates are currently: (1) Eu-ropean Space Agency (ESA) and (2) Max-Planck-Gesellschaft (MPG, Germany).

Present ASTRONET membership repre-sents about 80 % of the total astronomicalresources in Europe. One of the goals of ASTRONET is to involve other coun-

tries who may wish to join, at the mostappropriate level. If necessary, a third levelof involvement will be introduced.

How will astronomers contribute?

Full participation from the European astro-nomical community at large is essential to ensure the quality and validity of theoutput, in particular of the Science Visionand the Infrastructure Roadmap. A num-ber of panels are being established, in thefirst case organised by (broad) scientificthemes, in the second case by observa-tional technique.

The panels will develop draft reports, tobe amended and validated through de-dicated symposia addressing the wholecommunity. These will be held aroundDecember 2006 for the Science Vision,and March 2008 for the Infrastructure

Roadmap. Careful attention will be givento avoid unnecessary duplication, e.g.through close contacts to other relevantFP6 activities like the ELT and SKA De-sign Studies. With ESA participating fullyin the exercise, an important objective is to achieve better coordination betweenthe plans for ground- and space-basedastronomy. For further information onASTRONET status, goals and objectives,see our web site at http://www.astronet-eu.org

ASTRONET will be a true success if ithelps to join the forces of all fundingagencies towards the development of thefuture, large and expensive research infra-structures that European astronomy willneed to keep abreast, and to develop truecollaborations between all European as-tronomical communities. The enthusiasticcontribution of the community in this ef-fort will be essential for reaching this goal.

The five work packages (deep blue)and their sub-tasks (light blue), with thecorresponding work package and task leaders in parentheses. The inter-action between the various tasks isshown schematically.

3. Infrastructure Roadmap (PPARC)

Long-term Astronet (INSU)

Forum with other FP6 I3 (PPARC)1. Networking (INAF)

New Participants (NOTSA)

2. Science Vision (ESO)

Cooperation Toolbox (PT-DESY)

Pilot Collaboration (NWO/NOVA)

Strategy & Resource Census (ESO)

Science Vision (NWO/NOVA)

5. Management (INSU)

Inter-Agency Coordination (INAF)

4. Coordinated Actions (PT-DESY)

Implementation (INSU)



The Current and Future ST-ECF

Robert A. E. Fosbury (ST-ECF)

After 15 years in orbit, the Hubble SpaceTelescope is facing an uncertain futurealthough there is a real possibility of a fifth servicing mission within the nexttwo to three years. ESA and ESO havejointly been pondering the future of the ST-ECF at the ESO headquarters inGarching and this article outlines theirconclusions.

The Space Telescope – European Coordi-nating Facility (ST-ECF) was establishedby ESA and ESO in 1984 for the purposeof supporting European astronomers intheir use of the Hubble Space Telescope(HST). The subsequent history of thegroup and its evolving role up to 2002 isdescribed in an article in the STScI News-letter (Fosbury and Albrecht 2002;http: //sco.stsci.edu/newsletter/PDF/2002/summer_02.pdf) and will not be repeatedin detail here.

With the HST already in orbit beyond itsdesign lifetime of fifteen years, both ESAand ESO have been giving thought to thefunctions that the ST-ECF should per-form during the remaining active phase ofthe Hubble project – however long that is – and also to their own requirements formaintaining such a joint group into thepost-Hubble era. The purpose of this arti-cle is to outline the results of these de-liberations and to sketch the anticipatedfuture activities of the group.

Current tasks

Following a mid-term review in 1996, car-ried out by Len Culhane, Rolf-PeterKudritzki and George Miley, the ST-ECFshifted emphasis from user support todirect contributions to the HST project incollaboration and coordination withSTScI. It also took on additional tasks andsome additional staff at the request ofESA as part of a renewed ESA/NASAHubble Memorandum of Understanding(MoU). These tasks were to pre-processthe plate scans used for the secondGuide Star Catalog (GSC II), to improvethe calibration of selected post-oper-ational Hubble instruments using instru-ment modelling techniques, and to pro-

vide a European channel for Hubble-relat-ed public outreach. The first of these has been completed, the second is beingabsorbed within the now primary ST-ECFtask of providing high-level science data products for selected current – aswell as post-operational – Hubble instru-ments. The available manpower has,however, been reduced as a result of thecurrent evolution of the ESA/NASA MoUwhich takes account of ESA’s consid-erable contributions to JWST. The third:European Outreach for Hubble, JWSTand some other ESA activities, is continu-ing as a mainstream ST-ECF activity.

By early 2006 and at the request of ESA,the ST-ECF will have reduced its stafffrom a maximum of twenty-one back tothe original complement of seven ESA-funded and seven ESO-funded personnelwith the expectation that this level will be maintained until the decommissioningof the HST spacecraft. When this eventwill happen is currently very uncertain. IfNASA decides to carry out the next serv-icing mission (SM4) before any cata-strophic failure, this could be many yearsahead: at least five years after the servic-ing mission takes place. However, forplanning purposes, ESA is using the dateof 2010 as the anticipated Hubble End ofLife (EoL).

While the HST remains operational – andprobably somewhat beyond EoL, the ST-ECF will continue its focus on Hubble-related tasks, principally in the areas of high-quality science-data products andpublic outreach. Projects currently under-way in the instrument science area arepredominantly in the area of spectrosco-py, notably the provision of high quality,physical-model-based wavelength, chargetransfer efficiency calibrations and cor-rection procedures for STIS and the cali-brations and procedures for reducing the slitless spectroscopy data from theACS. The STIS project is a natural follow-on from similar work done for FOS andthe ACS work has its heritage in NICMOSand STIS. Descriptions of this work canbe found in recent issues of the ST-ECFNewsletter which can be found on-line atthe ST-ECF website (http://www.stecf.org/newsletter/ ). The wavelength calibrationwork has involved extensive laboratorywork to characterise the on-board lampsand has resulted in a new set of wave-

lengths for the Pt/Cr-Ne lamp (Sansonetti,C. J. et al. 2004).

If SM4 is carried out successfully withinthe next two to three years, Hubble will be equipped with two new instruments: aNUV-NIR imager (WFC3) and a high-ef-ficiency UV spectrometer (COS). There isalso the realistic possibility that the STIS will be repaired. Such a dramatic in-crease in HST capabilities would carryserious implications for the ground sup-port at both the STScI and at the ST-ECF.

It has been clearly apparent that there are a number of resonances betweenthese Hubble-based instrument projectsand the interests of ESO for several VLT instruments. The physical similaritiesbetween STIS and UVES have alreadybeen exploited in the current UVES pipe-line and the similar problems facingCRIRES are currently yielding to similarlyeffective procedures for the high-qualitycalibration of this new infrared spectrome-ter. There has been a recent feasibilitystudy for the application of the slitlessspectroscopy procedures developed forACS to the MXU-mode of the FORS2spectrometer; work that is described in the article by Kuntschner et al. on page19 of this issue of The Messenger.

The fruits of these HST instrument sci-ence labours are ultimately destined forthe so-called Hubble Legacy Archive(HLA) which has been conceived to bethe repository of the highest-quality,science-ready data products that can beconstructed by the combined efforts of the STScI, the Canadian AstronomicalData Centre (CADC) and the ST-ECF.Destined to be fully VO-compliant, theHLA will – together with the publishedscientific papers based on the data (cur-rently somewhat in excess of 5 000) – be the lasting legacy of this extraordinarilysuccessful scientific endeavour.

The public are overwhelmingly supportiveof the HST observatory and, at least inthe USA, the word “Hubble” has becomesynonymous with astronomical telescope.The efforts to make the European publicmore aware of the project and ESA’s rolein it are led by the small (2.5 FTE) ST-ECFoutreach group who has set up a veryefficient and productive infrastructure toincrease public awareness – especially


amongst young people. The recent cele-bration of Hubble’s 15th year in orbit has had a very high worldwide visibilitywith more than half a million copies of the DVD “HUBBLE – 15 years of discov-ery” being distributed. The web site(http://www.spacetelescope.org/ ) hasbecome one of the most visited sciencesites in the world.

Future role

In the post-Hubble era, both ESA andESO will continue to share common inter-ests and goals. Given their common ob-ligations to astronomy and the very largeoverlap between the communities theyserve, both organisations have expresseda desire to maintain a joint ESA/ESO ac-tivity based at the ESO Headquarters in Garching. This group will carry out arange of technical, outreach and coor-dination tasks of interest and benefit toboth parties.

The current plans for such a group, whichwould evolve from the ST-ECF, have beendeveloped with ESA and ESO over thelast year or two. They include a continua-tion of the high-quality science-data prod-uct intitiative which necessarily forms the foundation of the VO endeavour. Tar-get projects would come from both ESAand ESO instrument developments. Inaddition, there are a number of coordina-tion activities between the two organisa-tions that would benefit from a joint groupin Garching. The first steps have alreadybeen taken with the publication of theESA/ESO Working Group report on extra-solar planets described in the previous is-sue of The Messenger. This will shortly be followed by a report on the synergiesbetween Herschel and ALMA and then by one on Fundamental Cosmology.These are anticipated to be the beginningof a continuing series of initiatives thatmap out the joint interests and capabilitiesof both ESA and ESO to address majorscientific problems.

Finally, the continuation of a vigorous pub-lic outreach programme is considered to be vital to ensure a continuing interestin the progress of science into the nextgeneration and the future ST-ECF willcarry out part of this programme for ESAin close collaboration with ESO.

As a footnote to this description, it shouldbe pointed out that the ST-ECF has been in the past and is currently a sub-stantial contributor to the ESO AstronomyFaculty and the scientific environment.With seven of its staff being Faculty mem-bers, it is well represented on Facultycommittees and currently holds the (elect-ed) Faculty chairmanship.

Reference

Sansonetti, C. J., Kerber, F., Reader, J., and Rosa, M. R., “Characterization of the Far-ultravioletSpectrum of Pt /Cr-Ne Hollow Cathode Lamps asUsed on the Space Telescope Imaging Spectro-graph on Board the Hubble Space Telescope”,2004, Astrophysical Journal Supplement Series153, 555–579

PPARC Council at ESO

On October 27, members of the PPARC Coun-cil paid a full-day visit to the ESO Headquar-ters. The Council members were welcomed byESO’s Director General, Dr. Catherine Cesarsky,and in the course of the day received an ex-tensive briefing on key ESO activities includingthe current science activities at ESO’s La SillaParanal Observatory in Chile, the status of the ALMA project and the OWL project for anextremely large optical/near-IR telescope.Furthermore, ESO’s instrumentation programme– arguably the world’s most comprehensiveand ambitious instrumentation developmentprogramme in astronomy – was described aswell as industrial relations and technologytransfer, and ESO’s substantial education andpublic outreach activities. The Council mem-bers also had an opportunity to see the ESOScience Data Archive and the Integration Hall where new instruments are tested prior tobeing shipped to the observing sites in Chile.During lunch, the participants had a chance to meet UK and Commonwealth staff workingat the Headquarters.

Pho

to:H

.H.H

eyer

,ES

O



Open House at ESO Garching

Henri Boffin (ESO)

On October 22, ESO in Garching openedits doors to the public, as part of thetraditional all-campus Open House. Theevent was successful again this year,clearly generating a lot of public interest,with 2000–2500 visitors passing throughour building within a seven-hour period.For ESO, the Open House is an importantelement of our continued dialogue withthe local community, playing a key role inmaintaining awareness about astronomyin general and about our organisation inparticular.

About forty ESO staff, from many differentdivisions, participated in the event to pro-vide, in an attractive and varied pro-gramme, information about ESO and theongoing work at this organisation. The

visitors could, for example, watch a spe-cially edited 15-min movie about ESO,listen to different talks on topics such as“exoplanets”, “the life of an astronomer”or ”spectroscopy”, view demonstrationsof image processing, the Virtual Ob-servatory, the principle of interferometry, CCD cameras and infrared detectors, as well as examine the VLT, ALMA andOWL models. Regular videoconferenceswere also done with Paranal and they allgenerated a lot of interest and enthusi-asm in the public. In addition, there wasan indoor planetarium – always very suc-cessful among children, an exhibition of astronomical paintings and ceramics as well as exhibitions of the Volksstern-warte München and the Observatory ofKönigsleiten. The Personnel Departmentprovided visitors with information aboutemployment opportunities at ESO, and

the ST-ECF also had a stand. The ESOamateur group, AGAPE, had kindly pre-pared several activities, viz. the obser-vation of the Sun and a demonstration ofradio astronomy.

The visitors were also given a chance toparticipate in the famous ESO Quiz, andmany entered the competition. Among thenumerous correct answers to the twelvequestions, 46 winners were selected andwere offered books, CD-Roms or posters.

Visitors came from all over Bavaria, butalso from Austria, and even from Italy and France. Many expressed their satis-faction with the activities being offered.There was clearly a lot to do and see assome even spent the whole day at ESO!The Open House was

also very success-ful among children.

The OWL model at-tracted much interest.

The videoconferenceswith Paranal generated a lot of interest and en-thusiasm.

Pho

tos:

H.H

.Hey

er,E

SO

(3)


Danish Minister visits ESO Chile

Gonzalo Argandoña, Felix Mirabel (ESO)

On October 21, the Danish Minister forEducation, Mr. Bertel Haarder, paid a visitto ESO in Santiago, accompanied by asmall entourage including the Danish Am-bassador to Chile, Anita Hugau.

Mr. Haarder, who had visited La Silla 15 years ago, was received by the ESORepresentative in Chile, Felix Mirabel, andClaus Madsen, Head of the ESO PublicAffairs Department. The guests were alsointroduced to two Danish fellows whocurrently work at ESO in Chile, Lise Chris-tensen and Thomas Dall.

During the meeting, the minister wasbriefed on the current science operationsin Chile, the relationship between ESOand Chile, the status of the ALMA projectas well as on the progress regardingESO’s ELT plans.

The discussion also covered ESO’s many-sided educational outreach activities, both in Chile and in Europe and howthese initiatives can play a role in a largereffort to stimulate interest in the naturalsciences amongst young people.

Pho

to:I

.Lem

us,E

SO

Pho

to:G

.Arg

ondo

ña,E

SO

At the ESO office in Santiago, BertelHaarder (Danish Minister of Educa-tion) talks to Lise Christensen. At the back, Claus Madsen, Felix Mirabeland Mrs. Haarder.

The Danish Minister was particularlyinterested in science educational proj-ects promoted by ESO in Chile andEurope. The picture shows children atthe 2005 Fair of Physics that took place in October in Santiago, as part ofthe celebrations of the Chilean NationalScience and Technology Week.

Almost five hundred science educatorsfrom twenty-seven countries met in No-vember at CERN, Geneva, for “Scienceon Stage” (SOS). The delegates had beenselected through elaborate national pro-cedures involving contests and national“festivals” in which participants were invit-ed to present their ideas and projects.

The SOS programme and indeed the Eu-ropean Science Teaching Festival is ar-ranged by the EIROforum, of which ESOis a member, and was developed from the successful “Physics on Stage” events.

An inspiring programme of presentationsand workshops, and a lively fair wheredelegates could demonstrate excitingexperiments and teaching projects, al-lowed educators to share techniques and take new ideas back to their schoolsacross Europe.

This dedicated initiative for teachers,which is co-funded by the EuropeanCommission, aims to promote innovativescience teaching in the formal educa-tion system. The event has gatheredincreasing support from education author-ities, learned societies, and industry in

Science on Stage 2005

Douglas Pierce-Price, Henri Boffin,Claus Madsen (ESO)

“Science on Stage”, the European Sci-ence Teaching Festival, is a major edu-cational outreach programme for sci-ence teachers. It aims to identify andfoster innovation within formal scienceeducation by means of exchange ofbest practice, workshops and seminarsinvolving educators from all overEurope.



many of the participating countries, whichare clearly keen to ensure that theireducational systems maximise the bene-fits from taking part in the programme.

Henri Boffin and Douglas Pierce-Pricefrom ESO presented workshops forteachers on the ALMA InterdisciplinaryTeaching Project and an introduction toGamma-Ray Bursts, respectively.

In another workshop, delegates learntabout and discussed the new Europeanjournal for science teachers, “Science in School”. This journal is based at theEMBL and published by the EIROforum. Itwill be launched in spring 2006 and pub-lished quarterly. It is part of the NUCLEUSproject funded by the European Commis-sion.

An international jury presented “EuropeanScience Teaching Awards” to the bestprojects at the fair, in a ceremony attend-ed by Jean-Michel Baer, the Director of Science and Society in the EC Directo-rate General for Research. The four gen-eral awards, together with individualawards from each of the seven EIRO-forum organisations, had a total value of 17000 € .

ESO at CER 2005

Claus Madsen (ESO)

On November 14–15, the Brussels Exhibi-tion Centre (Heysel) was the home of amajor international conference on sciencecommunication with the title “Communi-cating European Research” (CER 2005).

The conference was officially opened withspeeches by Commissioner Dr. JanezPotocnik (Commissioner for Science andResearch) and Commissioner VivianeReding (Commissioner for InformationSociety and Media), with former Commis-

ESO’s prize was awarded to the “EinsteinYear Library Project”, by Mandy Curtis,from the United Kingdom. The gener-al prizes were won by Catherine Garcia-Maisonnier of France for “Building aWeather Balloon at School” (1st prize),Wim Peeters of Belgium for “Physics is Cool! – the Box of Experiments” (2ndprize), Jerzy Jarosz and Aneta Szczgielskaof Poland for “The Cardiovascular Sys-tem” (3rd prize), and Tobias Kirschbaumof Germany for “Tracing Earthquakes(Chinese Seismograph)” (4th prize). The

Festival participantsenjoy the magic ofchemistry on stage.


Pho

to:H

.Bof

fin,E

SO

other EIROforum prizes went to MariaJoao Carvalho of Portugal for “Lichen andWater Quality” (CERN), Eilish McLoughlinet al. of Ireland for “Teaching Science as a Process” (EFDA/JET), EvanthiaPapanikolau of Greece for “DNA Helix”(EMBL), Agota Lang of Hungary for “Neu-rode, or Garfield the Lazy Cat” (ESA),Gianluca Farisi of Italy for “Humanism andScience” (ESRF), and Nanna Kristensen ofDenmark for “Jewellery is chemistry” (ILL).

The next Science on Stage event will takeplace in Grenoble in April 2007.

sioner Philippe Busquin, now a promi-nent member of the European Parliament,chairing the session.

Over the two days, about 2100 partici-pants from 56 countries, including morethan 200 journalists, discussed all as-pects of public science communication,including science education, informal science learning, science festivals andmedia work. The conference also saw the launch of the “Communiqué Initiative”,a first step towards creating a EuropeanMedia Centre for science.

ESO participated through the EIROforumpartnership, that acted as organisers of two sessions – one on formal scienceeducation and one on media work. ESAand ESO were also represented in a lively panel discussion on communicationof astronomy and space science. Finally,EIROforum had a major information standin the exhibition area.

The conference offered ample opportunityfor stimulating exchanges, not just be-tween European scientists and media ex-perts but also with participants fromoverseas, including the US and China. It


EU Research Commis-sioner Dr. Janez Potocnik(right) visiting the EIRO-forum stand in the CERexhibition hall.

Pho

to:M

.Ped

ouss

aut,

ES

A seems clear that public awareness andunderstanding of science is not a “side-issue” vis-à-vis scientific progress. Rather,it is increasingly seen as central to the fu-ture of science in a democratic society.Thus it is a burning issue across the entireworld and one that must be addressedthrough a large number of initiatives andon an appropriate scale.

Science Perspectives for 3D SpectroscopyReport on the Conference on

Jeremy Walsh (ST-ECF)Markus Kissler-Patig (ESO)

About four years ago when discussionswere taking place to plan the proposal to the European Commission for aResearch Training Network (RTN) on 3Dspectroscopy, we decided to make aninternational conference one of the clos-ing highpoints of the network. At that timethere were only a few 3D instrumentsroutinely taking data on large telescopes(such as Integral on the WHT and Oasison the CFHT) and some of us thoughtthat a full-scale international conferenceon science with 3D spectroscopy mightbe rather difficult to fill. However, as ittranspired, we had problems containingthe conference in four and a half days.The RTN, called Euro3D, shared the host-ing of the conference with ESO and it was held at ESO Headquarters in Garch-ing from October 10–14, 2005.

At the inception of the Euro3D RTN, inJuly 2002, it was perceived that Europe

had a strong instrument advantage inoptical and NIR 3D spectroscopy. By 3Dspectroscopy is generally meant thetechnique of obtaining multiple spectraover a 2D field of view; there are vari-ous implementations from scanning slits,to imaging Fabry-Perots, to integral field units (IFU) with fibres, lenslet arraysor slicers. A number of instruments werein the planning stage, not least threeplanned for the VLT – the IFU mode ofVIMOS, the FLAMES/Giraffe Argus modeand SINFONI. However, the expertise inhandling of the data was mostly confinedto the instrument groups themselves and there was a perceived “difficulty” inreduction and analysis of optical/NIR 3D data. This is partly a result of the largequantities of data delivered by IFU instru-ments but also a reflection that signifi-cant development beyond the tools forlongslit spectroscopy is required to ana-lyse the resulting data cubes (2 spatial + 1 spectral dimension). One of the aimsof the Euro3D RTN was to narrow thebridge of difficulty by training young re-searchers in 3D spectroscopy observation

Figure 1: The Euro3D Research Train-ing Network (www.aip.de/Euro3D)initiated and sponsored this confer-ence, together with ESO.

and analysis, so that they could spreadthe word that 3D data need not be intimi-dating. The RTN also planned softwaredevelopment, science projects, a confer-ence and an IAC winter school.

The RTN funding will be completed asplanned at the end of this year and hasproved a great success. There were ten young post-doc researcher positionsspread across ten European institutes all with connections to 3D spectroscopy(AIP (Potsdam), Cambridge, Durham, IAC(Teneriffe), IAP (Paris), Leiden, Lyon,Marseille, Milan, MPE (Garching)); therewas also a team from ESO but without a post-doc. Despite worries that it would



Figure 2: Kinematically Decoupled Cores (KDCs) arebest viewed with IFUs. Some, however, require adaptive optics systems for their resolution as shownhere in kinematic maps from OASIS, as comparedwith SAURON (taken from Richard McDermid’s pres-entation).

be difficult to find enough keen post-docswith interest and/or experience in 3Dastronomy, all the positions were occu-pied and two early departures were alsofilled. Most of the young researchers have moved (or are about to move) on tofurther positions in astronomy. Six of theEuro3D post-docs presented their workat the meeting, so the aim of showcasingtheir work as a culmination to the RTNwas satisfied.

The conference was aimed at 3D science,as there had been a previous technicalworkshop at Durham University in July(Integral Field Spectroscopy – Techniquesand Data Reduction), which was inciden-tally also partly sponsored by Euro3D. At the Garching conference there were atotal of 92 participants; with eight invitedtalks, 41 contributed talks and 30 posters(previewed by two poster oral sessions); it was a full schedule. However, there wastime for a visit to the Aying brewery late one afternoon, where much beer wastasted, followed by the conference dinner,Bavarian style.

The first afternoon was devoted to anoverview of current and future instrumen-tation to set the scene for the followingscience presentations. Guy Monnet (ESO)brought his many years of involvement in3D spectroscopy to the fore in the open-ing review on the past and up to thepresent. This was followed by a review byJeremy Allington-Smith (Durham) oncurrent instrumentation. Then there werea number of talks on instrument projectsin various states of planning and funding.Martin Roth (AIP, Potsdam) described the MUSE project with its 1; square fieldand 24 spectrographs, funded for theVLT. This instrument was out-multiplexedby HETDEX, the Hobby-Eberly TelescopeDark Energy EXperiment, with its pro-posed 145 identically replicated IFU spec-trographs described by Karl Gebhardt(University of Texas). While the sciencecase for MUSE is centred on deepsearches for high-z galaxies, HETDEX,with some initial funding for a prototype,aims to measure redshifts of one milliongalaxies to constrain the equation of stateof dark energy. SINFONI, the most re-cent 3D instrument to go into regular op-erations, had an exciting entry with FrankEisenhauer’s overview of the large paletteof impressive first scientific results.

The science covered at the conferencewas intendend to be wide and in fact cov-ered Solar System objects, extra-solarplanets, Galactic stars and nebulae, thestructure of nearby galaxies, and dis-tant galaxies out to the highest knownredshifts. Santiago Arribas (IAC) de-scribed IFU observations of the planetarytransit of HD 209458b which came closeto the spectrophotometric accuracy of HST observations and Jean-PierreMaillard (IAP) described applications ofimaging Fourier transform spectrometry,a technique which has been eclipsed in popularity by IFU’s in recent years.Most of the second day of the conferencewas devoted to the most mature field inwhich 3D spectroscopy has been applied– that of the morphology, kinematics and study of stellar populations in nearbygalaxies. The Sauron survey (conductedwith a dedicated instrument on the WHT) featured very prominently and EricEmsellem (Lyon) reviewed the field and extensively used an acronym new toat least some of us: KDCs – KinematicallyDecoupled Cores. It was clear that suchfeatures could not be guessed at from the

surface brightness distribution and couldbe lost if a slit was placed along the majoror minor axis of a galaxy only. The Sauronsurvey which has taken as a sample 72 galaxies with a representative range ofproperties (Hubble type, luminosity, loca-tion) is able to give statistical trends in 2Dgalaxy kinematic and chemical properties.

IFU observations are not only useful forwell-resolved nearby galaxies, but, as described by Lutz Wisotzki (AIP, Pots-dam), they are also well suited to smallermore irregular objects such as gravita-tional lenses. Both the lenses and thelensing galaxy can be imaged at once andspectral variations among the lenses,which can be caused by microlensing,can be simultaneously determined. In par-ticular the strong lensing by galaxy clus-ters creates curved sources which are notwell matched to long-slit spectra. IFU ob-servations in galaxy clusters are uniquelypowerful both for studying the clustergalaxies and the lenses. If the lensingmodel of the cluster is well determined,then the elongated lenses can be de-distorted to derive the true velocity field

–20 –10 0 10 20–4 –2 0 2 4

–30

30

km/s

–60

60

km/s

arcs

ec

–4

–2

0

2

4

–20

–10

0

10

20

arcs

ec

arcsecarcsec

OASISSAURON

Figure 3: Gravitational lenses recently became targetsof IFU observations. Atmospheric dispersion, whichshifts the position of the targets as a function of wave-length, can be used to advantage to provide spa-tial “super-sampling” of the IFU images (PMAS data,taken from Lutz Wisotzki’s presentation, see alsoWisotzki et al. 2003).

HE 0435-1223 λ = 4278 Å λ = 5601 Å λ = 6924 Å

Walsh J., Kissler-Patig M., Science Perspectives for 3D Spectroscopy


for background galaxies. A number of talks emphasised this approach (MarkSwinbank, Durham, and Marie Lemoine-Busserolle, Cambridge/Oxford); initialresults in the cluster RCS0224-002 werereported by Swinbank et al. in the last issue of The Messenger (121, 33).

IFUs do not always have to have a largefield: if they are targeted at a number ofsmall objects within one field then multipleIFUs are a more efficient use of preciousdetector pixels. This mode is alreadyavailable in FLAMES and observations ofthe velocity fields of z ~ 0.5 galaxies wasdescribed by François Hammer (Obser-vatoire de Paris). The ability to build upsamples of galaxies with resolved velocityfields at substantial redshift allows theevolution of the Tully-Fisher relation (ratherthe lack of evolution over the last ~ 4 Gyr)as also illustrated by near-IR IFU observa-tions with CIRPASS by Andrew Bunker(Exeter). Future multiple IFU instrumentsfor the VLT, such as KMOS, will explorethis area in the infrared.

The nuclei of galaxies with a rich array ofkinematic processes and strong gradientsin surface spectra present ideal targets for IFUs. Reinhard Genzel (MPE) updatedthe participants on the continuing saga ofthe Milky Way Galactic Centre, whereintegral-field observations have played afundamental role in discovery, of course in the IR on account of the high extinc-tion. As well as providing the kinematicsand stellar classifications of the lumi-nous young stars orbiting the centralblack hole, IFU observations, most latelytaken with SINFONI on the VLT, have

16 T 16 spectra0.19 e –17 rms

4400–5.0 T 10–17

5.0 T 10 –17

1.0 T 10 –16

1.5 T 10 –16

2.0 T 10 –16

3600 3800 4000 4200

0

[O II] map

NGC 6720PMAS guider image

4

5

6

8

Figure 4: The faint halos of planetary nebulae profitfrom the power of IFUs to integrate over a large aperture (without degrading the spectral resolution).The spectrum of a single spaxel (spatial element –top) is compared to the spectrum obtained with thesame area 16 u 16 spaxel binned (bottom) (takenfrom Martin Roth’s presentation).

–300

2.1 2.15

–200–100

0100200

Hβ [O III] 4959Å [O III] 5007 Å

VVDS 02029855z = 3.29

SINFONI with adaptive optics

[O III] velocity map

2.73 e – 17 rms

[O II] map

NGC 6720PMAS guider image

4400–5.0 T 10 –17

5.0 T 10 –17

1.0 T 10 –16

1.5 T 10 –16

2.0 T 10 –16

3600 3800 4000 4200

0

4

5

6

8

1 spectrum

Figure 5: SINFONI study of the galaxy VVDS 02029855 at a redshift z = 3.29. The CFHT I-band image shows the faint object, for whichSINFONI (assisted by adaptive optics in this case)obtained a spectrum. A velocity field was ob-tained from the detected [O III] 5007 Å line (taken from Marie Lemoine-Busserolle’s presentation).


Catherine Cesarsky, Bruno Leibundgut(ESO)

The scientific legacy of Alvio Renzini wascelebrated with a one-day workshop inGarching on October 19. Alvio retired asthe VLT Programme Scientist in June thisyear and this conference was a way tolook back at his many scientific achieve-ments and his influence on ESO and theVLT project.

Report on the

Science Day in Honour of Alvio Renzini


The programme of the day was only apartial reflection of Alvio’s scientific work.There were no presentations of his earlywork, e.g. on stellar evolution. Instead thetopics concentrated on research he iscurrently interested in and actively work-ing. The programme was as varied asAlvio’s work. The diversity was clear rightfrom the start: Pascale Jablonka (Lau-sanne) described work on the bulge ofM31, followed by a status report on white dwarfs in globular clusters given bySabine Moehler (ESO). Gianpaolo Piotto

(Padova) introduced globular cluster oddi-ties and Markus Kissler-Patig (ESO) usedglobular clusters as template stellar pop-ulations. A topic that was hot when Alviojoined ESO were flares at the centre ofelliptical galaxies, and Francesco Bertola(Padova) gave an update of this research.Moving into more cosmological topicsAdriano Fontana (Rome) talked about thegalaxy mass function at high redshifts andClaudia Maraston (Oxford) presentedwork inspired by Alvio on the importanceof AGB stars in the interpretation of

enabled the spectra of the central objectSag A* to be followed as it undergoesoutbursts on a time scale of hours. It wassalutory to see spectra of tens of stars inan area the size of a ground-based see-ing disc! The “weather” around the nearbySeyfert nucleus of NGC 1068 wasdescribed by Gerald Cecil (University ofNorth Carolina) with some fine anima-tions. This environment has been system-atically studied by IFUs and probes howshocks play an important role proximateto the active nucleus.

Emission-line nebulae were often con-sidered the staple targets for 3D instru-ments providing well-resolved, kine-matically complex targets. A number oftalks described imaging spectroscopystudies of planetary nebulae (Martin Roth,AIP, using the PMAS instrument; KatrinaExter, IAC, using Integral) and H II regionssuch as the proplyds in Orion (HenriPlana, LATO, Brasil, using the GeminiGMOS IFU) and Herbig Haro objects(Rosário Lopez, Universidad de Barcelo-na). The ability to post-bin the IFU dataenables the detection of very low surfacebrightness features, such as the faintemission-line haloes around planetarynebulae. Several talks described workdone with the MPFS on the Russian 6-mtelescope which has been collecting data

for almost 15 years. Sergei Fabrika(Special Astrophysical Observatory, Rus-sia) and Pavel Abolmasov (Moscow State University) discussed IFU observa-tions of the shock nebulae associatedwith Ultra-luminous X-ray binaries (highluminosity cousins of SS433); here He II

emission is an important diagnostic. The SNR around SN1987A is now largeenough for AO corrected imaging spec-troscopy and Karina Kjær (ESO) de-scribed VLT SINFONI observations.

Far from being restricted to large extend-ed objects, IFUs are beginning to makean impact in high-redshift observations to which the entire last day was dedicated.The often complex, knotty appearance ofhigh-redshift galaxies, which are probablyinteracting or merging, can be sampled in an unbiased way with an IFU. This canbe important since the obvious nucleusmay not be apparent and rotation curvescan be derived for apparently rather cha-otic objects and emission-line haloes, as shown for example by Montse Villar-Martin (Andalucia) and Richard Wilman(Durham). Deep surveys of blank sky, or regions around QSOs for indications oflocal overdensity of galaxies, are beingconducted with Sauron and the VIMOSIFU, as shown in several talks (see Jarvis et al. in the last issue of The Mes-

senger (121, 38)). Serendipitous discoveryof emission-line objects which can bemissed by photometric surveys was high-lighted, although the number of such ob-jects is small.

The conference summary was presentedby Andreas Quirrenbach (Leiden). He saidthat October 14 was a historic day: it wasthe last day of the last conference devot-ed to the subject of 3D spectroscopy. Thetechnique has come of age and can nowenter the repertoire of standard observa-tional tools employed by an astronomer.(Imagine a conference on long-slit spec-troscopy!). The very wide range of topicscovered at the conference to which imag-ing spectroscopy had been applied wasindeed striking. There seemed to be littlearea of the parameter space of spatialcoverage, spatial scale, spectral coverageand spectral resolution which had notbeen described at the meeting. The dis-cussion following the summary concen-trated on instrumental issues and possibledevelopments, such as the ideal detec-tor able to determine the position, wave-length and polarisation state of each in-coming photon, currently imperfectlyrealised but the ultimate goal of 3D spec-troscopy.


Pho

to:H

.H.H

eyer

,ES

Ospectral energy distributions at high red-shifts. The large surveys, for which Alvioplayed a pivotal role in having ESOmaking significant contributions were pre-sented by Piero Rosati (ESO), who gave astatus report on GOODS, and Simon Lilly (Zurich) provided a glimpse at what isto come for zCOSMOS. The latest updateon the continuation of the K20 survey and the GMASS project was presented byAndrea Cimatti (Arcetri). Massimo Tarenghi(ESO) looked back at the time when theVLT was being built and elaborated on theimportant scientific input and guidanceAlvio provided. The day was rounded ofby Alvio himself giving his view on cur-rent research (adding the latest result on asearch for eclipsing planets in the direc-tion of the Galactic bulge) and some morespeculative ideas. A recurring theme in all presentations was that Alvio must notretire from his scientific career and severalpleas were made for him to continue to contribute his physical insights into themany diverse projects he has been in-volved in.

Coming to ESO was a gamble for Alvio(maybe also for ESO). Entering the worldof hard observational realities as a theoristhas not always been easy. Even beforecoming to ESO, Alvio was deeply involvedin observational research. However, he was collaborating with observers andhelping them interpret their data. At ESOhe formulated important parts of the ESO programme, introduced ESO to sur-vey work and gave scientific direction for the use of the telescopes and futureinstrumentation. He sensed interestingscientific developments very early on andprepared ESO for them. He led a work-ing group to discuss the best ways to ob-serve extra-solar planets, which led to theconstruction of HARPS, and organisedthe best way to observe GRBs with theVLT. The latter has provided an observa-tional mode, the rapid response mode,which is not offered at any other ground-based observatory. The VLT has alwaysbeen at the heart of his concerns and hehas been an unwavering champion in itssupport.

Unknown to many astronomers Alvio hada defining influence on the way the VLT has developed and is used by astron-omers. He wrote the Level 1 Require-ments for the VLT kind of as a “warm-up.”

The VLT Science Policy document wasalso part of his work. Alvio led a group ofESO scientists to discuss conceptualtests of the VLT system (and define refer-ence programmes) to see whether thesystem was able to cope with them. Heeffectively assessed design and construc-tion decisions for their implications for the (then future) science applications andfound ways to involve the ESO sciencecommunity with the VLT project. Once thefirst telescope became reality Alvio led the science verification of the first UT andlater the instruments as they came online.

Alvio was deeply involved in the instru-mentation programme for the VLT. Heconstantly tried to improve the perform-ance of the instruments and made manyproposals for upgrades. The improve-ment of the red sensitivity of FORS2 washis proposal, as was the fibre connectionfrom FLAMES to UVES. He was toldseveral times that this was not possibleand should not be considered. But he persisted and today this is one of themost successful instrument combinations.

Alvio made sure that the second HDFhappened and was located in the South-ern Hemisphere so that the VLT wouldhave access. He followed through withthe large cosmological surveys and in-volved ESO in them. The ESO contribu-tion to GOODS was initiated by Alvio. TheISAAC IR imaging and the FORS2 (seethe article by Vanzella et al. on page 25 inthis issue of The Messenger) and VIMOSspectroscopy are due to his initiative and his constant support. zCOSMOS isthe logical continuation of these activities.

Alvio in general was a great ambassadorfor the VLT project in the astronomicalcommunity in Europe. He advocated VLTscience wherever he went.

As Alvio’s last months and weeks at ESOwere approaching many young peoplewere wondering “How will research atESO continue without Alvio?”. This ques-tion was asked many times at coffee and in the corridor. Who will replace Alvioand his scientific input, his scientific in-sights and his experience? This was (andis) a real concern among many youngerscientists who enjoyed the quiet scienti-fic leadership Alvio provided on the fourthand fifth floor of the ESO building in

Garching. His contributions to the journalclubs, seminars and colloquia were very much appreciated and they alwaysshowed a new and different aspect of the scientific topic at hand. His vast re-search experience was a guiding light formany at ESO. It was not by chance that he organised five ESO workshopsand provided new ideas every year.

Alvio has been amongst the most highlyregarded and respected ESO astrono-mers in the community. In a sense he rep-resented the scientific conscience of ESO.

The science day was a great success andenjoyed by all. We wish Alvio many moresuccessful and fruitful years to come.


The purpose of these summer schools isto provide the opportunity for young re-searchers to gain practical experience inobservational techniques, data reduc-tion and analysis and the use of virtualobservatory tools. Students will carry outsmall research projects, centred on se-lected astrophysical topics, in smallgroups under the supervision of experi-enced astronomers. These practical ex-ercises will be complemented by lectureson general observational techniques and archival research for both ground andspace based astronomy.

The observing school at the telescopes inHaute-Provence will largely concentrateon the skills required to execute an ob-serving programme (imaging and spec-troscopy), giving emphasis to spectrosco-

NEON Observing Schools

Announcements

The Fifth NEON Observing SchoolJuly 23–August 6, 2006Observatoire de Haute-Provence, France

The Second NEON Archive Observing SchoolAugust 30–September 9, 2006ESO, Germany

py and instrumental developments. TheArchive Observing School at ESO willconcentrate more on the quality appraisalof existing data and the research pos-sible with large databases combiningground and space data, with emphasis on data-reduction techniques and thenew tools now available within the VirtualObservatory.

The schools are sponsored by the Euro-pean Community, Marie Curie Actions.

The organising consortium is composedof Asiago Observatory (Italy), Calar AltoObservatory (Germany, Spain), ESO (Ger-many; including participation from theSpace Telescope European CoordinatingFacility), Observatoire de Haute-Provence(France), Institut d’Astrophysique de Paris

(France) and La Palma Observatories (INGand NOT, the United Kingdom, theNetherlands, Spain and Nordic Countries).

The schools are principally open to As-tronomy PhD students and postdocs whoare nationals of a Member State or anAssociate State of the European Union.Applications from non-European studentswill also be considered, depending on theresources available.

The application deadline is April 30, 2006for both schools.

For further details, see:http://www.eso.org/neon-2006 and http://www.iap.fr/eas/neonNew.html

March 6–10, 2006, Universidad de Concepción, Chile

The Universidad de Concepción, the Cen-tro de Astrofísica FONDAP (Chile), theCerro Tololo Interamerican Observatoryand ESO are organising a conference on “Globular Clusters – Guides to Galax-ies”, which will take place on March 6–10,2006.

The principal question of whether andhow globular clusters can lead to a betterunderstanding of galaxy formation andevolution is perhaps the main driving forcebehind the overall endeavour of study-ing globular-cluster systems. Naturally, thisdisperses into many individual problems.

The objective of our conference is to bringtogether researchers, both observationaland theoretical, to present and discussthe most recent results. Topics to be cov-ered are: internal dynamics of globularclusters and interaction with host galaxies(tidal tails, evolution of cluster masses),accretion of globular clusters, detailed de-scriptions of nearby cluster systems,ultracompact dwarfs, formations of mas-sive clusters in mergers and elsewhere,the ACS Virgo survey, galaxy formationand globular clusters, dynamics and kin-ematics of globular-cluster systems, anddark-matter-related problems.

Invited speakers: Holger Baumgardt,Michael Beasley, Kenji Bekki, Jean Brodie,Andreas Burkert, Rupali Chandar, PatrickCôté, William Harris, Michael Hilker,Andres Jordan, Oleg Gnedin, DeanMcLaughlin, Bryan Miller, Kathy Perrett,Aaron Romanowsky, François Schweizer,Stephen Zepf.

Scientific Organising Committee: BruceElmegreen, Duncan Forbes, Doug Geisler,Eva Grebel, Leopoldo Infante, MarkusKissler-Patig, Søren S. Larsen (co-chair),Tom Richtler (chair).

Visit our webpage: www.astro-udec.clContact: [email protected]

Globular Clusters – Guides to GalaxiesConference on


Personnel Movements

Arrivals

Europe

Arnaboldi, Magda (I) User Support AstronomerAspinall, Gareth (GB) Project PlannerBoneva, Kristina (BG) StudentCaproni, Alessandro (I) Software EngineerDi Lieto, Nicola (I) Control EngineerDoherty, Michelle (AUS) FellowEsteves, Raul (P) Electronics EngineerJost, Andreas (D) Electronics EngineerKainulainen, Jouni (FIN) StudentKotak, Rubina (EAK) FellowKurlandczyk, Hervé (F) EngineerMartinez, Patrice (F) StudentMöhler, Sabine (D) Operations ScientistNeeser, Mark (D) Operations ScientistParker, Laura Catherine (CDN) FellowPierce-Price, Douglas (GB) Education OfficerSivertsen, Beatrice (D) SecretarySwat, Arkadiusz (PL) Optical EngineerValenti, Stefano (I) StudentVandame, Benoît (F) Software EngineerVillegas Mansilla, Daniela (RCH) Student

Chile

Christensen, Lise Bech (DK) FellowDe Figueiredo Melo, Claudio (BR) Operations AstronomerDurand, Yves (F) Head of the Engineering Department

ParanalGeissler, Kerstin (D) StudentGilmour, Rachel Emily (GB) User Support AstronomerJames, Gaël (F) FellowJullo, Eric (F) StudentNoterdaeme, Pasquier (B) StudentRisacher, Christophe (F) Instrument ScientistSchütz, Oliver (D) Operations AstronomerVehoff, Stefan (D) Student

Departures

Europe

Ahmadia, Aron (USA) StudentAlmagro Garcia, Susana (E) SecretaryBlondin, Stéphane (F) StudentCarmona Gonzalez, Andres (CO) StudentKniazev, Alexei (RUS) User Support AstronomerMignani, Roberto (I) Operations ScientistMottini, Marta (I) StudentNylund, Matti (S) Software EngineerSilva, David Richard (USA) Head Data Flow Operations

Chile

Amado Gonzalez, Pedro Jose (E) Operations AstronomerBaumont, Sylvain (F) StudentDelle Luche, Céline (F) StudentGil, Carla (P) StudentHousen, Nico (B) Software EngineerLombardi, Gianluca (I) StudentNesvacil, Nicole (A) Student

October 1, 2005–December 31, 2005

List of Proceedings from the ESO Astrophysics Symposia

Title

Growing Black Holes: Accretion in a Cosmological Context

High Resolution Infrared Spectroscopy in Astronomy

Multiwavelength Mapping of Galaxy Formation and Evolution

Science with Adaptive Optics

Astronomy, Cosmology and Fundamental Physics

Toward an International Virtual Observatory

Extragalactic Globular Cluster Systems

From Twilight to Highlight: The Physics of Supernovae

The Mass of Galaxies at Low and High Redshift

Lighthouses of the Universe: The Most Luminous Celestial Objectsand Their Use for Cosmology

Scientific Drivers for ESO Future VLT/VLTI Instrumentation

The Origin of Stars and Planets: The VLT View

Gamma-Ray Bursts in the Afterglow Era

Deep Fields

Mining the Sky

Editors

Andrea Merloni, Sergei Nayakshin, Rashid A. Sunyaev

Hans-Ulrich Käufl, Ralf Siebenmorgen, Alan F. M. Moorwood

Alvio Renzini, Ralf Bender

Wolfgang Brandner, Markus Kasper

Peter A. Shaver, Luigi DiLella, Alvaro Giménez

Peter J. Quinn, Krzysztof M. Górski

Markus Kissler-Patig

Wolfgang Hillebrandt, Bruno Leibundgut

Ralf Bender, Alvio Renzini

Marat Gilfanov, Rashid A. Sunyaev, Eugene Churazov

Jacqueline Bergeron, Guy Monnet

João F. Alves, Mark J. McCaughrean

Enrico Costa, Filippo Frontera, Jens Hjorth

Stefano Cristiani, Alvio Renzini, Robert E. Williams

Anthony J. Banday, Saleem Zaroubi, Matthias Bartelmann

Volume

2005

2005

2005

2005

16/2003

15/2004

14/2003

13/2003

12/2003

11/2003

10/2003

9/2003

8/2003

7/2003

6/2003

Contents

Telescopes and InstrumentationR. Sharples et al. – Surveying the High-Redshift Universe with KMOS 2S. D’Odorico – Instrument Concepts for the OWL Telescope 6The Centre of the Active Galaxy NGC 1097 9L. Pasquini et al. – CODEX: Measuring the Expansion of the Universe 10Afterglows of Elusive Short Gamma-Ray Bursts 14T. Wilson – ALMA News 15ALMA Antenna Contract Signed 17G. Argandoña, F. Mirabel – Inauguration of the APEX Telescope 18H. Kuntschner et al. – Towards an Automatic Reduction of

FORS2-MXU Spectroscopy 19P. Padovani, P. Quinn – The Virtual Observatory in Europe and at ESO 22

Reports from ObserversE. Vanzella et al. – GOODS’ Look at Galaxies in the Young Universe 25L. J. Tacconi et al. – The Dynamics and Evolution of

Luminous Galaxy Mergers: ISAAC Spectroscopy of ULIRGs 28Supernova in NGC 1559 31P. E. Nissen et al. – Lithium Isotopic Abundances in Metal-Poor Stars 32C. Evans et al. – The VLT-FLAMES Survey of Massive Stars 36M. P. Döllinger et al. – Why are G and K Giants Radial Velocity Variables? 39

Other Astronomical NewsA.-M. Lagrange – ASTRONET: Towards a Strategic Plan

for European Astronomy 42R. A. E. Fosbury – The Current and Future ST-ECF 44PPARC Council at ESO 45H. Boffin – Open House at ESO Garching 46G. Argandoña, F. Mirabel – Danish Minister visits ESO Chile 47D. Pierce-Price, H. Boffin, C. Madsen – Science on Stage 2005 47C. Madsen – ESO at CER 2005 48J. Walsh, M. Kissler-Patig – Report on the Conference on

Science Perspectives for 3D Spectroscopy 49C. Cesarsky, B. Leibundgut – Report on the Science Day

in Honour of Alvio Renzini 52

AnnouncementsConference on Globular Clusters – Guide to Galaxies 54NEON Observing Schools 54Personnel Movements 55List of Proceedings from the ESO Astrophysics Symposia 55

ESO is the European Organisation forAstronomical Research in the SouthernHemisphere. Whilst the Headquarters(comprising the scientific, technical andadministrative centre of the organisa-tion) are located in Garching nearMunich, Germany, ESO operates threeobservational sites in the Chilean Ata-cama desert. The Very Large Telescope(VLT), is located on Paranal, a 2 600 mhigh mountain south of Antofagasta. AtLa Silla, 600 km north of Santiago deChile at 2 400 m altitude, ESO operatesseveral medium-sized optical tele-scopes. The third site is the 5 000 mhigh Llano de Chajnantor, near SanPedro de Atacama. Here a new submil-limetre telescope (APEX) is in opera-tion, and a giant array of 12-m submil-limetre antennas (ALMA) is underdevelopment. Over 1600 proposals aremade each year for the use of the ESOtelescopes.

The ESO MESSENGER is published fourtimes a year: normally in March, June,September and December. ESO alsopublishes Conference Proceedings andother material connected to its activities.Press Releases inform the media aboutparticular events. For further infor-mation, contact the ESO Public AffairsDepartment at the following address:

ESO HeadquartersKarl-Schwarzschild-Straße 285748 Garching bei MünchenGermanyPhone +49 89 320 06-0Fax +49 89 320 23 [email protected]

The ESO Messenger:Editor: Peter ShaverTechnical editor: Jutta Boxheimerwww.eso.org/messenger/

Printed by Peschke DruckSchatzbogen 3581805 MünchenGermany

© ESO 2005ISSN 0722-6691

Front Cover Picture: H II region NGC 1982 in the Orion Nebula ComplexThis image of the H II region NGC 1982 (M43) was obtained from observations madewith the WFI instrument on the 2.2-m telescope at La Silla in December 2001 byMassimo Robberto and colleagues. The dramatic rendering of this image is due to a combination of three narrow band filters (O III, Hα and S II r, with a total of 1.5 hoursof integration). The observations were reduced and combined by Benoît Vandame(ESO).

the messenger 122 · 2010. 3. 24. · pickoff module one of the more unusual kmos elements is the...

Documents