Installation and Commissioning of FLAMES, The VLT Multifibre Facility
Installation and Commissioning of FLAMES, The VLT Multifibre Facility
Introduction
1
Table 1. Observational capabilities of FLAMES.
*The GIRAFFE resolving power (R) and wavelength coverage given here are only average values. They will vary for the different set-ups by ± ~ 20%. Aperture
is in arcseconds.
July 3. We report here on the complex (www.eso.org/instruments/FLAMES/ pling of 0.3 arcsec/micro-lens and a to-
process of integration and commission- manuals). tal coverage of 6.6 by 4.2 arcsec. 15
ing, and we compare the expected and The characteristics of the different deployable ARGUS single sky fibres
observed astronomical requirements. observing modes are given in Table 1, are also available.
which summarizes the observational GIRAFFE is operated with 30 fixed
Overview capabilities of FLAMES. set-ups (22 high-resolution + 8 low-res-
olution modes). For performance esti-
FLAMES is the multi-object, interme- UVES-Fibre mates (based on expected transmis-
diate and high-resolution fibre facility of sion curves and performances) the
the VLT. Mounted at the Nasmyth A UVES is the high-resolution spectro- user is referred to the Exposure Time
platform of UT2 it offers a large correct- graph of the VLT UT2 (D’Odorico et al. Calculator (ETC), available at the ESO
ed field of view (25 arcmin diameter) 2000). Each positioner plate has eight ETC web page: www.eso.org/observ-
and it consists of several components, fibres connected to the UVES Red Arm. ing/etc
developed by several consortia (Aus- With an aperture on the sky of 1 arcsec, The FLAMES observing software
tralis (AAO/ANU/UNSW), Paris-Meu- the fibres project to five UVES pixels (OS) is coordinating the operations of
don (OPM), Geneva and Lausanne giving a resolving power of ~ 47,000. the Telescope, the Positioner and the
(OGL), Bologna, Cagliari, Palermo and Only the three standard UVES Red set- UVES and GIRAFFE spectrographs. In
Trieste (ITAL Consortium)) and ESO: ups are offered, with central wave- addition, it allows COMBINED observa-
• An optical Corrector, providing ex- lengths of 520, 580 and 860 nm, re- tions: that is the simultaneous acquisi-
cellent image quality and telecentricity spectively. tion of UVES and GIRAFFE spectra
over the full field of view of 25 arcmin with the specific observing modes listed
diameter (developed at ESO). GIRAFFE in Table 1.
• A Fibre Positioner hosting two
plates. While one plate is observing, GIRAFFE is a medium-high resolu- LAYOUT
the other one is positioning the fibres tion spectrograph (R = 6000–33,000)
for the subsequent observations, there- for the entire visible range (370–950 Figure 1 shows a view of the main
fore limiting the dead time between ob- nm). It is equipped with two echelles for components of the FLAMES facility, the
servations (developed at AAO). low and high resolution and uses inter- fibre Positioner and GIRAFFE, as seen
• A link to the UVES spectrograph ference order sorting filters to select the on the telescope platform. The Cor-
(Red Arm) via eight single object fibres / required spectral range within an order. rector support is also visible, attached
plate (developed at OPM in collabora- The typical spectral coverage in one at the Nasmyth rotator.
tion with ESO). exposure is 60–100 nm in low resolu-
• A high and intermediate resolution tion and 20–40 nm in high resolution. 1. Corrector
optical spectrograph, GIRAFFE (devel- The fibre system feeding GIRAFFE
oped at OPM and ESO). consists of the following components The optical corrector is a big doublet
• Three types of fibre systems: MEDU- (cf. Avila et al. 2002 for details): of 880 mm free aperture. The function
SA, IFU, ARGUS (developed at OPM). • MEDUSA fibre slits, one per posi- of the corrector is to give an excellent
• A coordinating observing software, tioner plate. Up to 132 separate objects image quality over the whole 25 arcmin
that allows simultaneous UVES and GI- (including sky fibres) are accessible in FLAMES field of view and to provide a
RAFFE observations (developed at ITAL MEDUSA single fibre mode, each with pupil located at the centre of curvature
Consortium in collaboration with ESO). an aperture of 1.2 arcsec on the sky. of the focal plate, to avoid vignetting for
• As for all VLT instruments, a full • IFU slits. 15 deployable Integral off-axis fibres. When the whole optical
Data Reduction Software (DRS) pack- Field Units (IFU) per plate, each con- train is taken into account (including tel-
age is provided, integrated in the ESO sisting of an array of 20 square micro- escope optics and vignetting), the ef-
Data Flow Software (developed at OGL lenses, for a total (almost rectangular) fective transmission of the corrector de-
with contributions from OPM (GI- aperture of ~ 3 × 2 arcsec. For each pends on the observing wavelength
RAFFE) and ITAL Consortium (UVES- plate there are also 15 IFU dedicated to and on the distance of the object to the
Fibre)). sky measurements. field centre, varying from about 78% at
The FLAMES components are de- • 1 ARGUS slit. A large integral unit 370 nm to 90% in the visible, and 82%
scribed in detail elsewhere (cf. e.g. Avi- consisting of a rectangular array of 22 at 1 micron.
la et al. 2002, Blecha et al. 2000, Gilling- by 14 micro-lenses, fixed at the centre A system of lamps has been installed
ham et al. 2000, 2002, Jocou et al. 2000, of one positioner plate. Two scales are to illuminate the Nasmyth screen,
Mulas et al. 2002, Pasquini et al. 2000, available: one with a sampling of 0.52 which is located in the telescope cen-
Royer et al. 2002), on the ESO web site arcsec/micro-lens and a total aperture trepiece, about 2 m in front of the focal
and in the FLAMES User Manual of 12 by 7 arcsec, and one with a sam- plane.
2
2. Fibre Positioner (OzPoz) ferent modes for
GIRAFFE. It is
The Fibre Positioner (OzPoz) is at worth mentioning
the core of the FLAMES facility. OzPoz that, in addition to
is a rather large and complex system the object or sky
equipped with four plates, two of which fibres coming from
are currently in use (cf. Gillingham et al. the Positioner,
2002). The Positioner can be subdivid- each GIRAFFE fi-
ed into the following subsystems: bre slit has five fi-
• Plates: Two metallic disks, on which bres devoted to si-
the magnetic buttons holding the fibres multaneous wave-
are attached. Each of the plates has a length calibration,
hole in the centre. In one plate this hole which provide si-
hosts ARGUS. Each plate has a curva- multaneous cali-
ture of 3950 mm to match the curvature bration spectra for
of the corrector focal surface. each observation
• Retractors: Mechanical systems acquired with GI- Figure 2: Transmission distribution of all GIRAFFE fibres, as meas-
maintaining the fibres in constant ten- RAFFE (Hammer ured in the laboratory.
sion. Each fibre is equipped with one et al. 1999).
retractor. The retractors are externally The histogram distribution of the 1.47 times the fibre diameter core,
the same for all fibres. transmission is given in Figure 2 for the which implies that the contamination
• Exchanger: Main structure holding different fibre types. The FLAMES fi- between adjacent fibres is about 10 per
the plates. The exchanger can perform bres have been proven very robust, so cent. With the chosen arrangement of
two main movements: it can retract (or that out of more than 1000 fibres, only the fibres this is acceptable since in
approach) the Nasmyth adapter to en- a few were broken (and replaced) at the normal seeing conditions a higher level
gage the plate (or disengage it). end of the whole process. of contamination will be present at the
Furthermore, it can rotate the upper fibre entrance level.
part in order to perform the plate ex- 3.1 UVES fibres
change. 3.4 ARGUS fibres
• r-θ system and Gripper: This unit Each of the positioner plates hosts
grips and releases the magnetic but- eight 55-metre fibres to guide the light The ARGUS system is a fixed array
tons at the positions reached via the r- to the UVES spectrograph located on of 14 × 22 micro-lenses, similar to the
θ (polar) robot. The gripper requires a the opposite Nasmyth Platform. From IFUs, located in the middle of Plate 2.
back-illumination system, sending a the UVES simultaneous calibration ARGUS is also equipped with a lens to
brief light flash along each fibre from box, one additional 5-metre fibre reach- switch between a scale of 0.52″/micro-
the spectrograph towards the plate. A es the UVES-Fibre slit. The fibre cen- lens to a finer scale of 0.3″/pixel. An
camera records this back-illumination tres are separated by 1.7 fibre core, ADC ensures that the object images at
light and performs an image analysis. implying that there is some degree of different wavelengths are maintained in
The back-illumination light is used for contamination between adjacent fibres; the same locations up to a Zenithal dis-
several purposes: to reach the required the UVES-Fibre Data Reduction Soft- tance of 60 degrees. In addition to the
positioning accuracy and to detect if the ware (Mulas et al. 2002) has been de- object fibres, fifteen ARGUS sky fibres
magnetic button was properly lifted veloped to obtain a complete de-blend- are present on the plate.
from the plate. ing of the spectra.
• OzPoz is equipped with a calibra- 4. GIRAFFE
tion box which can direct the light either 3.2 MEDUSA fibres
from a tungsten, or from a Th-Ar, or GIRAFFE is a fully dioptric spectro-
from a Ne lamp. In this way FF, Th-Ar Each plate also hosts 132 MEDUSA graph. The fibres are arranged in 5 long
and Ne calibrations can be obtained for fibres. The separation between each of slits (IFU 1/2, MEDUSA 1/2, ARGUS) at
GIRAFFE and for UVES. the MEDUSA fibres in a sub-slit is 2.26 the curved focal plane of the collimator.
• Field Acquisition Bundles (FACBs): × the fibre core; this ensures a fibre-to- One of these fibre slits is placed in the
Four magnetic buttons on each plate fibre contamination below 0.5 per cent. working position by a translation stage.
are equipped with a system of 19 co- During our tests we have experi- The other slits are masked. After leav-
herent fibres each. This bundle is used enced some (about 2%) increased ing the fibres at F/5, the light first pass-
to obtain images of fiducial stars, one stray light due to reflections from the fil- es through one order sorting filter be-
per button. The four images are viewed ter and the (polished) exit slit. We will
by an ESO technical CCD; the image attempt to reduce this with a stray light
centroids are computed and the proper mask that reduces the width of the re-
offsets are calculated to centre the fidu- flecting exit slit viewed from the side of
cial stars into the bundles. Each FACB the filter.
bundle has an effective diameter of 2.4
arcseconds. 3.3 IFU fibres
• Positioning Software: It is written
around the so-called delta-task pro- Each Integral Field Unit (IFU) button
gram, developed initially for the 2dF is composed of twenty micro-lenses
system at AAO. This program allows arranged in a rectangular shape. The
crossing between the fibres and it de- micro-lenses are 0.52″ squares. The
termines the button movements se- movable IFUs are a unique characteris-
quence. tic of GIRAFFE and can be placed all
over the FLAMES field of view, with the
3. Buttons and Fibre Systems exception of a small fraction of the cen-
tre annulus. Each plate hosts fifteen
FLAMES is equipped with different IFUs plus fifteen Sky IFUs. The sepa- Figure 3: Transmission of order sorting filter
types of fibres for UVES and for the dif- ration between the fibres’ centre is only HR 17.
3
Table 2. Summary of GIRAFFE construction characteristics. lines/mm coated with MgF2/Al. Its first-
order blaze lies at 1.96 µm, so in the
Fibre slit height 76.8 mm range 370–950 nm it works in orders
Type Echelle + Order selection Filters 5–2; eight set-ups are needed to cover
Collimator aperture F/5 the full range. After being dispersed,
Collimator beam 18 cm the light passes again through the colli-
H-R Echelle 204 × 408 mm, 316 lines/mm, 63.4º Blaze mator, forms a real image at an inter-
L-R Echelle 156 × 204 mm, 600 lines/mm, 33.7º Blaze mediate focal plane and is finally im-
Reimager demagnification ratio 2.5 aged by an F/2 reimaging camera with
Effective Camera Focal Length 360 mm 7 elements.
Detector 2048 × 4096, 15 µm EEV CCD
Scale 0.20 arcsec/pixel 4.4 Spectral Format and Efficiency
Figure 4: Component efficiencies per set-up and the resulting overall DQE (slit to detected photoelectrons) of GIRAFFE. All reported values
are averages; note that grating efficiencies can vary by up to a factor of two within a set-up. The optics efficiency includes vignetting at the
grating, this is why the reported LR values are slightly lower than HR values at the same wavelengths.
4
adapter were encountered. These
problems were cured in advance of the
August commissioning run. On July 3,
first GIRAFFE spectra could be ob-
tained. Two sets of spectra of stars in
an astrometric field were acquired,
each of 10 minutes. The set-up used
was LR4, Low Resolution centred at a
wavelength close to the V filter. 66
MEDUSA fibres were positioned (some
in sky positions), and the reduced spec-
tra of some of them are shown in Figure
7. The stars have R magnitudes be-
tween 13 and 14. In all spectra the
Magnesium triplet at 518 nm is clearly
visible, as well as the 557 sky emission
Figure 5: Th-A spectrum taken with a MEDUSA slit. The fibres in the centre of the slit have
line.
lines moved towards smaller pixels; e.g. fibres corresponding to the central fibres have slight- The stability and repeatability of GI-
ly redder wavelength coverage than the ones at the CCD edges. RAFFE on the telescope, measured by
taking a ThAr spectrum every 4 hours
over a period of 2 weeks is shown in
Figures 8 and 9.
The tests have been very satisfac-
tory, showing that the instrument ex-
ceeds specifications.
A further commissioning took place in
August 2002, enabling us to advance
the commissioning tests. Many
MEDUSA ‘rasters’ have been per-
formed, to determine the telescope-
plate geometry.
In these rasters an astrometric field is
observed with MEDUSA and the tele-
scope is offset along a grid around the
pointing position. For every star, the
distance between its nominal position
(the centre of the grid) and the positions
of its maximum flux is then computed
and used to correct the astrometric
Figure 6: Reduced spectra of 7 giants in the globular cluster Omega Cen taken with UVES- model. The results have been so far
fibre around the Hα line. very encouraging, and the found resid-
uals are shown in Figure 10 , with an
During the first run at the end of clearly not a suitable permanent solu- RMS of 0.15 arcseconds, which is very
March 2002 the Positioner was inte- tion. This was corrected prior to a sec- good when considering that this num-
grated and tested and we could gather ond commissioning period in June but ber includes all sources of errors, e.g.,
the first light with UVES-Fibres. Three new mechanical problems, related to the accuracy of the stellar astrometry.
10-minute spectra were obtained for 7 the interface with the telescope rotator- In Figure 11 we show the image of 4
stars (one fibre was dedicated to sky)
belonging to the Globular cluster
Omega Cen. Three of them were previ-
ously observed with UVES in long slit
mode (Pancino et al. 2002). A small
portion of the spectra around the line
Hα line is shown in Figure 6. The stars
had magnitude varying between V =
11.5 and 12.9. After data reduction the
equivalent widths of a few hundred
lines were compared for the UVES –
Fibre and UVES-long slit spectra of the
same star, finding an excellent agree-
ment: EQW(Fibre) = EQW(Slit) + A with
A in the range of 4–6 mÅ, with an rms
of 4–7 mÅ, which is largely due to
equivalent widths measurement uncer- Figure 7: Reduced
tainties. spectra of GIRAFFE
During this March run, malfunctions from the first com-
were experienced in the positioner ro- missioning run: the
stars belong to an
bot, due to the lower temperature at the astrometric field.
telescope compared to those prevailing The Mg triplet and
during the tests in Australia and in the the 557 nm sky
Assembly Hall in Paranal. We tem- emission line are
porarily fixed this using three heaters, clearly visible in all
spectra.
5
Figure 8: the average position of ThAr lines in a time series that was
taken at 4-h intervals over a period of 2 weeks. Shifts are given in
pixels relative to the first exposure. These exposures were taken in
Medusa HR mode, with configuration changes between exposures.
The jump on May 7 cannot be correlated to any known event like an
Figure 10: Residuals vectors after the subtraction of the position cen-
earthquake.
troid from the object fibre, after a successful modelling of the plate
geometry.
6
Figure 12: S/N ratio as measured in clean
regions of spectra taken in the L8 set-up,
compared with the Poisson S/N ratio ex-
pected on the number of counts recorded.
7
Figure15: GIRAFFE Low (L4, black lines) and High (H9, red lines)
resolution spectra of 4 giants belonging to the Globular Cluster NGC
6809
8
(Garching and Paranal) and at the dif- P. Molaro, R. Pallavicini, I. Porceddu Hammer, F., Hill, V., Cayatte, V. 1999, Jour-
ferent institutions who have contributed (ITAL Consortium). nal des Astronomes Français, 60, 19.
to the realization of this ambitious proj- Jocou, l. et al. 2000, Proc. SPIE Vol. 4008,
ect: J. Alonso, P. Ballester, J.-L. Be- References p. 475.
Mulas, G., Modigliani, A., Porceddu, I.,
ckers, P. Biereichel, B. Buzzoni, C. Ca- Avila, G. Guinouard, I., Jocou, L., Guillon, F., Damiani, F. 2002, in Instrument Design
vadore, N. Cretton, C. Cumani, B. Dela- Balsamo, F. 2002, in Instrument Design and Performances for Optical/Infrared
bre, A.V. Kesteren, H. Kotzlowski, J.-L. and Performance for Optical/Infrared Ground-Based Telescopes, Proc. SPIE
Lizon, R.S. Moreau , W. Nees, R. Pal- Ground-Based Telescopes Proc. SPIE 4841 (in press).
sa, E, Pozna, (ESO), P. Barriga, R. 4841 (in press). Pancino, E. et al. 2002: ApJ 568, L101.
Castillo, J. Spyromilio AND MANY Blecha, A., North, P., Cayatte, V., Royer, F., Si- Pasquini, L. et al. 2000, Proc. SPIE Vol.
OTHERS at Paranal, C. Evans, R. mond, G. 2000, Proc. SPIE Vol. 4008, p. 467. 4008, p. 129.
Haynes, U. Klauser, B. Hingley, D. D’Odorico, S. et al. 2000, Proc. SPIE Vol. Pasquini , L. et al. 2002, in Instrument
Mayfield, S. Miziarski, R. Muller, W. 4005, p. 121. Design and Performances for Optical/
Gillingham, P., Miziarski, S., Klauser, U. Infrared Ground-Based Telescopes, Proc.
Saunders, G. Schafer, K. Shortridge, 2000, Proc. SPIE Vol. 4008, p. 914. SPIE 4841 (in press).
(AAO), L. Jocou and the fibre team, T. Gillingham P.R., Popovic D., Waller L.G., Royer, F., Blecha, A., North, P., Simond, G.,
Melse and the mechanics workshop Farrell T.J. 2002, in Instrument Design Baratchart, S., Cayatte, V., Chemin, L.,
team, S. Baratchart, L. Chemin, H. Flo- and Performance for Optical / Infrared Palsa, R. 2002, in Astronomical Data
res, D. Horville, J.M., Huet, F. Rigaud, Ground-Based Telescopes, Proc. SPIE Analysis II, Proc SPIE 4847 (in press).
F. Sayède (GEPI, OPM), F.R. Ferraro, 4841 (in press).
with the collaboration of the HARPS Team: G. AVILA3, P. BALLESTER 3, J.-L. BECKERS 3, W. BENZ 5,
J.-L. BERTAUX 6, F. BOUCHY 1, B. BUZZONI 3, C. CAVADORE 3, S. DEIRIES 3, H. DEKKER 3, B. DELABRE 3,
S. D’ODORICO 3, W. ECKERT 2, J. FISCHER 5, M. FLEURY 1, M. GEORGE 1, A. GILLIOTTE 2, D. GOJAK 2,3,
J.-C. GUZMAN 2, F. KOCH 3, D. KOHLER 4, H. KOTZLOWSKI, D. LACROIX 3, J. LE MERRER10, J.-L. LIZON 3,
G. LO CURTO 2, A. LONGINOTTI 3, D. MEGEVAND1, L. PASQUINI 3, P. PETITPAS 4, M. PICHARD1, D. QUELOZ1,
J. REYES 3, P. RICHAUD 4, J.-P. SIVAN 4, D. SOSNOWSKA1, R. SOTO 2, S. UDRY 1, E. URETA2, A. VAN
KESTEREN 3, L. WEBER1, U. WEILENMANN 2, A. WICENEC 3, G. WIELAND 3 and advice by the Instrument
Science Team: J. CHRISTENSEN-DALSGAARD 7, D. DRAVINS 8, A. HATZES 9, M. KÜRSTER9, F. PARESCE 3,
A. PENNY 11
1Observatoirede Genève; 2ESO, La Silla; 3ESO, Garching; 4Observatoire de Haute Provence;
5Physikalisches Institut, Bern; 6Service d’Aéronomie du CNRS; 7Aarhus University, 8Lund Observatory,
9Tautenburger Landessternwarte; 10Laboratoire d’Astrophysique de Marseille; 11Rutherford Appleton Laboratory
Introduction Institut der Universität Bern (CH) and course also be offered to the astronom-
Service d’Aéronomie du CNRS (F) un- ical community like any other ESO in-
An extensive review of past, present der the leadership of the Observatoire strument.
and future research on extrasolar plan- de Genève (CH) where the Principal
ets is given in the article “Extrasolar Investigator and the Project Office are The Radial Velocity Method:
Planets” by N. Santos et al. in the pres- located. Further contributions come Error Sources
ent issue of The Messenger. Here we from ESO-La Silla and ESO Head-
want to mention only that the search for quarters (Garching). The agreement A description of the radial velocity
extrasolar planets and the interpreta- was finally signed in August 2000. method for the detection of extrasolar
tion of the scientific results have According to this agreement, the planets in general and of the CORALIE
evolved in recent years into one of the HARPS Consortium bears the cost for spectrograph installed at the Swiss
most exciting and dynamic research the spectrograph and all its compo- Leonhard Euler Telescope at La Silla in
topics in modern astronomy. nents whereas ESO provides the particular was given by Queloz & Mayor
As a consequence, ESO decided to Cassegrain Fibre Adapter, the fibre link, (2001); it describes the technique ap-
have a dedicated spectrograph built to the dedicated HARPS room in the tele- plied also with HARPS and the spec-
be installed at the La Silla 3.6-m tele- scope building and the complete detec- trograph which is its direct ancestor. An
scope. An Announcement of Oppor- tor system. In return, the HARPS instrument like CORALIE achieves an
tunity was issued in 1998 and a con- Consortium will be granted 100 HARPS accuracy in radial velocity determina-
sortium chosen for the realization of the observing nights per year for a period of tion of about 3 m/s, and with this accu-
project which consists of Observatoire 5 years after successful Provisional racy planets down to a minimum mass
de Haute Provence (F), Physikalisches Acceptance in Chile. HARPS will of of about one third the mass of Saturn
9
have been discovered. For HARPS an again according to formula (1), higher smaller than 1/1000 of a CCD pixel. A
accuracy of 1 m/s RMS is specified. To precision is obtained when the spectral temperature variation by 1°C or the
achieve this goal, we have to analyse resolution R is increased. Unfortu- change of atmospheric pressure by
the limitations that are encountered by nately, telescope diameter D, spectral 1 mbar would already produce effects
a spectrograph in order to overcome resolution R, and fibre diameter φ (pro- of the order of 100 ms–1! We have
them. jected on the sky) are not independent, therefore decided to act on different
but linked by the formula fronts:
• Stabilize the spectrograph environ-
Photon noise ment. HARPS is installed in vacuum to
, (2)
remove effects caused by varying am-
The contribution of photon noise to bient pressure. In addition, the temper-
the radial-velocity (RV) measurement where h is the diameter of the collimat- ature of the whole instrument is active-
error can be expressed, according to ed beam on the echelle grating and tan ly controlled.
Hatzes & Cochran (1992), by the β is the tangent of the echelle grating • Simplify the opto-mechanical de-
generic formula angle. In other words, if we want to in- sign, avoid any moving or movable
crease the spectral resolution we have components, design a robust spectro-
σ RV ∝ S –0.5 · λλ –0.5 · R –1.5. (1) to increase the collimated beam diame- graph.
ter, or restrict the field of view φ and • Use a spectral reference to track
The measurement precision is pro- consequently decrease the slit effi- residual instrumental drifts.
portional to the square root of the flux S ciency. Since it is impossible to avoid drifts of
and the wavelength range λλ of the A trade-off therefore had to be made 10 nm on the instrument, the spectral
spectrograph. It should be mentioned between spectrograph size, spectral reference is mandatory. On HARPS we
that in practice it is not the wavelength resolution, and slit efficiency. Our solu- will have the possibility of using either a
range that determines the precision, tion consists in building an instrument ThAr spectral lamp in simultaneous ref-
but rather the number of spectral lines using the largest monolithic echelle erence mode, or an iodine absorption
and their depth, since the information grating available to date (a 837 × 208 cell in self-calibrating mode. Both tech-
on RV is contained in the stellar ab- mm grating developed for UVES) and niques attain to date similar perform-
sorption lines. Nevertheless, the formu- to balance spectral resolution against ance at the level of about 3 ms–1. The
la tells us that we must maximize the slit efficiency. The compromise led to a main mode of HARPS will be the ThAr
amount of light entering the spectro- spectral resolution of R = 93,000 which simultaneous reference, since it allows
graph. This goal can be reached by in- in most cases is sufficient to resolve the to cover a spectral range 3 times larger
creasing the spectral range, maximiz- stellar absorption lines. Going beyond than with the iodine cell, which in addi-
ing the optical efficiency of the instru- would not reduce the photon noise but tion absorbs about 50% of the light. The
ment, and, of course, by using the only increase slit losses. At the select- efficiency is thus 6 times higher using
largest telescope available. ed resolution the fibre covers 1 arcsec the ThAr simultaneous reference
The UVES-like optical design as well on the sky. The resulting slit losses are mode.
as the choice of the optical components about 50% on average. It should be re- Often, stability of the instrumental
make HARPS an efficient spectro- called however that the goal was to op- profile (IP) is associated with the RV
graph. However, this is not sufficient to timize the instrument for RV precision, precision of an instrument. In practice,
get a precise RV spectrograph, since, which is proportional to R1.5 but only both spectral reference methods allow
proportional to the to avoid errors produced by internal IP
square root of S. variations. It is however true that the
Thus, a fibre twice ThAr simultaneous reference technique
as large would does not account for variations of the
collect twice the slit illumination. This technique works
amount of light, but therefore only if combined with a fibre
would also reduce feed and with an image scrambler,
the spectral resolu- whose sole function is to “scramble” the
tion by a factor of 2. light entering the fibre at the telescope.
The resulting RV The goal is to avoid a change in spec-
precision would be trograph illumination when the position
reduced by a fac- of the star on the fibre entrance
tor 2. changes. We estimate that the residual
effects are of the order of 0.25 ms–1 and
Instrumental can thus be neglected. It is worth men-
errors tioning that good telescope guiding and
high spectral resolution reduce these
If we want to de- kinds of error sources even further.
tect RV variations
of 1 ms–1 we must External errors
be able to detect
changes in the stel- We should not forget that RV meas-
lar absorption line urements may be influenced by error
position which are sources related neither to the instru-
ment nor to the telescope. Any of those
can produce RV errors of the order of
several ms–1 if not correctly accounted
Figure 1: The for. Without going into detail, we men-
finished HARPS tion the following:
Cassegrain Fibre
• Atmospheric absorption lines which
Adapter ready to be
shipped from La Silla vary in relative position and intensity
to Geneva for compared to the stellar spectrum
system tests. • Atmospheric dispersion
10
• Sunlight reflected by the moon and
superimposed on the stellar spectrum
• Stellar companions close to the tar-
get star may contaminate its spectrum,
depending on seeing conditions
• Stellar jitter, intrinsic variability of
the star
11
is thus 40962 pixels (15 µm square) of
which a field of 62.7 × 61.4 mm is actu-
ally used at a sampling of 4 pixels per
spectral element. The performance of
the chips is summarized in Table 2.
As HARPS is a stationary instrument,
its detectors are cooled by a continuous-
flow cryostat (CFC) of the current ESO
standard design. A special feature of the
HARPS cryostat is however the fact that
the detector head (visible on the left in
Fig. 4) is mounted to the spectrograph
bench inside the vacuum vessel, with the
actual CFC outside. Both are connect-
ed by a stainless steel bellows which
protects the detector high vacuum
(10–6 mbar) from the mere “emptiness”
(10–2 mbar) of the spectrograph vessel.
The detector head window also serves
as field lens of the camera optics.
As a consequence of this configura-
tion, all cables connecting the detector
head to the FIERA controller have to
pass through vacuum feed troughs.
Figure 3: The open HARPS vacuum vessel. The back side of the echelle grating is visible on Thermal enclosure
top of the optical bench as well as the rail system used for moving the large parts of the vessel.
In order to keep the spectrograph
echelle grating, a copy of the UVES erence fibres which is reflected off the temperature as constant as possible it
mosaic, is operated in quasi-Littrow con- gap on the echelle between the two was decided to put the vacuum vessel
dition. An f/2.1 parabolic mirror serves gratings comprising the mosaic. in an additional thermal enclosure. This
as collimator and is used in triple pass. is a well insulated room on the coudé
A dioptric camera images the cross-dis- Detector system floor of the 3.6-m telescope building
persed spectra (one each from the ob- which is itself already temperature sta-
ject and reference fibres) side by side HARPS employs a mosaic of two bilized. Based on preliminary measure-
onto a mosaic of two 2k × 4k EEV EEV type 44-82 CCDs (nicknamed ments we expect to keep the tempera-
CCDs. 68 orders cover a spectral range Jasmin and Linda). The spectral format ture variations of the spectrograph
from 380–690 nm. At a spectral resolu-
tion RS of 90,000, determined by the fi-
bre diameter, a spectral resolution ele-
ment is sampled by 4 pixels. All optical
components are mounted on a stain-
less steel optical bench. The optical pa-
rameters are listed in Table 1.
Procurement of the optics turned out
to be the most demanding of all. The
first company charged with the produc-
tion of the cross disperser grism failed
to deliver and a new contract had to be
negotiated with another supplier. The
collimator mirror also failed to material-
ize, both within schedule and with ac-
ceptable quality. The mirror which was
finally delivered had to be re-polished
by another company to meet our re-
quirements.
In order to improve the observing ef-
ficiency by always applying the correct
exposure time (according to the select-
ed signal-to-noise ratio SNR) we fitted
an exposure meter, again following the
example of UVES. Two photon coun- Figure 4: The figure shows the CCD detector system mounted below the spectrograph bench.
ters are used to separately measure Visible are also the field lens and a holder which still connects the head and the CFC. In the
the light coming from the object and ref- final configuration the CFC will be mounted outside the vacuum vessel.
System 2 fibres (1″ dia.), spectral range 380–690 nm, collimated beam diameter 208 mm
Echelle R4, 31.6 gr/mm, blaze angle 75°, mosaic 2 ×1 on Zerodur monolith 840 × 214 × 125 mm, efficiency > 65% in the visible
Cross disperser FK5 grism, 257.17 gr/mm blazed at 480 nm, 240 × 230 × 50mm, T = 73% (av)
Collimator Zerodur, f = 1560 mm, used diameter 730 mm, triple pass
Camera All dioptric, 6 elements in 6 groups, f = 728 mm, f/3.3, T > 85%
12
Table 2: Properties of the HARPS detector system. Mayor (2001)), we have estimated the
RV precision of HARPS as a function of
DQE 82–85% peak, 70% minimum stellar magnitude (see their Fig. 14).
Read-out noise ~ 3e¯ at 50 kpx/sec (2 ports/chip) The total RV efficiency of HARPS is
CTE 0.999999 (horizontal +vertical) at 166 kpx/sec thus about 75 times higher than that of
Cosmetics Science grade (grade 1) CORALIE. This extraordinary RV effi-
Read-out modes 50 kpx/sec and 625 kpx/sec, one or two ports per chip ciency will offer exciting new possibili-
ties in other research fields, for exam-
ple in asteroseismology.
bench inside the vacuum vessel (the sponse of the grating. The angle of the
only place that really matters) below echelle grating was not centred correct- Operations
0.1K, ideally at 0.01K! ly at that time thus delivering an asym-
metric blaze response. This was cor- Often neglected, observational effi-
rected later by tilting the echelle grating ciency also plays an important role. In
Performance
physically by 0.25°. fact, searching for exoplanets requires
The total efficiency of the instrument that many measurements per object
The instrument
has not yet been determined. This are carried out. Telescope time must
Recently we have completed the in- measurement will be done during therefore be used efficiently.
tegration of HARPS at the Geneva Commissioning using a reference star. From the beginning, strong emphasis
Observatory. Tests of the complete Nevertheless, the optical components was put on efficient operations scenar-
spectrograph have been carried out, were measured individually. Table 3 ios to optimize the yield from an ob-
and the spectroscopic characteristics summarizes the expected optical effi- serving run. Most important is certainly
verified. The spectral format recorded ciency of HARPS derived from these the adoption of the VLT Data Flow Sys-
corresponds well to the calculated val- measurements. From this, and based tem, from the preparation of Observa-
ues. All the orders – except order 115 on data recorded with the CORALIE in- tion Blocks with P2PP through the use
which falls between the two CCDs – strument at La Silla (see Queloz & of BOB at the telescope to the imple-
could be localized and extracted using
a tungsten flat-field lamp (Fig. 5), while
the wavelength calibration was done
using the ThAr spectral lamp. In Figure
6 we show a small segment of the ex-
tracted and wavelength-calibrated ThAr
spectrum. About 3500 spectral lines
covering a large intensity range could
be detected and identified. From the
spectrum shown, which was recorded
during an exposure of 7 s, we estimate
that the internal photon noise on the ra-
dial velocity is only 10 cm s–1. This rep-
resents a large gain compared to
CORALIE, in part due to the higher
optical efficiency of the spectrograph,
but mainly due to the higher spectral
resolution. Indeed, the Gaussian fit of a
single ThAr line indicates a spectral
resolution of R = 98,000, which is about
5% higher than expected.
As mentioned above, HARPS is
equipped with an iodine cell for the self-
calibration mode. We have recorded a
spectrum of our iodine cell by sending
the white light of the tungsten lamp
through it. The extracted and wave-
length-calibrated spectrum is shown
in Figure 7. The intensity of the con-
tinuum varies strongly across the
echelle order because of the blaze re-
Tel. + atm. 44% 47% 54% 57% 59% 59% 61% 63%
“Slit” 46% 47% 48% 49% 50% 50% 51% 52%
Instrument 8% 11% 15% 15% 17% 17% 15% 12%
CCD 65% 78% 85% 85% 81% 79% 76% 72%
13
matically reduced in near real time by a
pipeline running on a dedicated work-
station: at the end of the night the ob-
server will have available all final radial
velocities, ready for publication! All data
can be immediately written to DVD-
ROM and are also archived in the cen-
tral Garching science archive.
Outlook
Acknowledgements
14
OmegaCAM: the 16k × 16k CCD Camera
for the VLT Survey Telescope
K. KUIJKEN 1,2, R. BENDER 3, E. CAPPELLARO 4, B. MUSCHIELOK 3, A. BARUFFOLO 5,
E. CASCONE 4, O. IWERT 6, W. MITSCH 3, H. NICKLAS 7, E.A. VALENTIJN 2, D. BAADE 6,
K.G. BEGEMAN 2, A. BORTOLUSSI 5, D. BOXHOORN 2, F. CHRISTEN 2,6, E.R. DEUL1,
C. GEIMER 6, L. GREGGIO 5, R. HARKE 7, R. HÄFNER 3, G. HESS 6, H.-J. HESS 3, U. HOPP 3,
I. ILIJEVSKI 3, G. KLINK 8, H. KRAVCAR 3, J. L. LIZON 6, C. E. MAGAGNA5, PH. MÜLLER 9,
R. NIEMECZEK 6, L. DE PIZZOL5, H. POSCHMANN 8, K. REIF 8, R. RENGELINK 1, J. REYES 6,
A. SILBER 6, W. WELLEM 7
1
Leiden Observatory; 2NOVA/Kapteyn Astronomical Institute, Groningen;
3
Universitäts-Sternwarte München; 4INAF - Osservatorio Astronomico di Capodimonte, Napoli;
5
INAF – Osservatorio Astronomico di Padova; 6ESO, Garching; 7Universitäts-Sternwarte Göttingen;
8
Sternwarte Bonn; 9Radioastronomisches Institut Bonn
15
the opening edge of the one blade and
the closing edge of the other blade with
an identical time difference, even if the
blades are still accelerating – this pro-
vides an impact-free, high-accuracy
photometric shutter. Tests of the shutter
confirm that it meets the key technical
specification: for an exposure time as
short as 1 second, deviations from a
homogeneous exposure are well below
±0.2% over the whole field of view.
The OmegaCAM control electronics
are based on VME-based local Control
Units (LCUs), with a higher level of
Unix-based workstations to manage
the user interface, coordination, testing
and maintenance. The LCU is a stand-
alone VME crate equipped with a
Motorola CPU board, an Ethernet
board, the real time operating system
VxWorks, as well as specialized con-
Figure 2: General drawing of the OmegaCAM instrument. The view is dominated by the cryo- trol and interface boards. All the con-
genic cooling system (with pre-amplifiers, vacuum equipment, etc. attached on the outside) trolled functions are standardized as
and the detector head on top of it containing the CCD mosaic. Above the CCDs, obscuring much as possible, and the modular de-
the curved dewar window, is the filter exchange system (shown in blue), which moves filters sign facilitates the maintenance and
between the two magazines and the beam. The exposure shutter is mounted at the top of the should ensure efficient and reliable op-
instrument, just below the telescope flange.
erations.
The mechanics are designed and
built by the German consortium part-
minute. The whole detector system is out, allowing efficient observing in spite ners in Göttingen (H. Nicklas – housing,
mounted behind a large, curved dewar of the rather large distance the filters filter exchanger) and Bonn (K. Reif –
window (the final optical element in the have to travel. shutter). More details can be found in
VST design) and is cooled using a 40-l The filters are large (in the language Nicklas et al. (2002) and Reif et al.
Nitrogen cryostat. Readout of the full of our latest ESO member state: a (2002). The control electronics were
mosaic takes 45 s, and is accomplished square foot) and heavy: when fully load- designed at INAF-Naples by E.
by two FIERA controllers (a third ed with 12 filters, the instrument will con- Cascone, and are being assembled at
FIERA takes care of the four guiding tain 40 kg (90 lbs.) of filter glass alone! Munich University Observatory under
and image analysis CCDs). The exposure shutter (Reif et al. W. Mitsch.
The OmegaCAM detector team at 2002) is one of the key units of
ESO is led by O. Iwert. OmegaCAM (Fig. 4). It consists of two 3.3 Optics
carbon fibre blades which open and
3.2 Hardware close the light path. They are driven by The VST telescope will work in two
In front of the dewar window is the micro-stepper motors and move configurations, which can be selected
mechanical part of OmegaCAM: clos- smoothly on linear motion guides. remotely. In the standard configuration,
est to the CCD window sits the filter ex- These movements are controlled such foreseen for work at small zenith dis-
change mechanism, and above that the that each individual CCD pixel ‘sees’ tances, a two-lens field corrector is
shutter. Both components have to fit
into a design space of a mere 16 cm
between the dewar window and the
VST’s Shack-Hartmann unit. The hous-
ing provides the mechanical link be-
tween the telescope flange and the de-
tector/cryostat system.
Figure 2 gives a section view of the
final design that foresees a cylindrical Figure 3: The
housing with a spoke-like rib structure 1.5-m diameter
to support the axisymmetrical loads at housing struc-
the Cassegrain focus. The housing can ture during a
test assembly of
be seen in Figure 3. the main units:
The filters are stored in two maga- the two storage
zines which can move up and down, magazines (to
either side of the focal plane, through be inserted into
large shafts in the housing. A linear the big shaft at
stage slides filters into the beam, where left and hidden
they are locked into place by means of right), the filter
movable notches. High-precision filter exchange unit
positioning ensures that intensity varia- with the
exchange car-
tions in the flat fields due to optical im- riage and a 420
perfections in the filters (dust grains, × 320 mm2
etc.) are less than 0.1%. The filter ex- opaque,
change unit is built in such a way that it aluminium ‘filter’
allows one filter to be pulled into the (at the bottom
beam while the previous one is pushed left).
16
used. The second configuration re-
places this corrector with one including
an Atmospheric Dispersion Corrector
(ADC), consisting of one lens and two
counter-rotating prism pairs. The oper-
ating wavelength ranges are 320–1014
nm and 365–1014 nm for the two-lens
corrector and corrector + ADC respec-
tively.
The only optical parts located in the
instrument are the filters, and the en-
trance window to the cryostat, which
doubles as a field lens.
The primary filter set of OmegaCAM
will be a set of Sloan u′, g′, r′, i′ and z′
filters. In addition, there will be Johnson
B and V filters for stellar work and for
cross-calibrating the photometric sys-
tems, a Strömgren v filter, an Hα filter
consisting of 4 segments with redshifts
of up to 10,000 km/sec, and a seg-
mented ugri filter for efficient photomet- Figure 4: The OmegaCAM exposure shutter. The aperture size is 370 × 292 mm, the short-
est possible exposure time is smaller than 1 msec, the deviations of the effective exposure
ric monitoring of the sky. time from pixel to pixel (homogeneity) are smaller than ± 0.2% for a 100-msec exposure, the
The procurement of large format fil- exposure time accuracy is about 0.3 msec. (The laptop computer gives an idea of the
ters of the required size turned into a scale.)
challenging task. Only one manufactur-
er (the French company SAGEM, for-
merly REOSC, who figured the VLT pri- of the readouts by the different maintain the instrument, not individual
mary mirrors) could make an offer for FIERA’s, and to the efficient storage of data sets, calibrated at all times. ESO
producing the primary set of filters with- the data on disk. will operate the instrument in service
out resorting to a segmented design, The Instrument Software is being mode, optimizing the observing pro-
which would have created vignetting produced by the Italian part of the con- gramme to ambient conditions, and
shadows on the detector array. Rather sortium, headed by A. Baruffolo (INAF- routinely taking calibration data. Thus
than using coloured glass – barely Padua), and is described in more detail each night the instrument’s overall re-
available in the required size – the filter in Baruffolo et al. (2002). sponsivity and also the transmission of
passband is generated by means of the atmosphere will be monitored in the
multiple layer coating of up to 5 sur- 4. Calibration and Data u′, g′, r′ and i′ bands irrespective of the
faces in a sandwich of three plates. The Reduction Software schedule of science observations. Data
expected throughputs of the Sloan fil- reduction recipes, run in ESO’s DFS,
ters are very high (Fig. 5). The amount of data produced by will provide a continuous characteriza-
Filter procurement is coordinated by OmegaCAM will be truly huge. We es- tion of the behaviour of the instrument
U. Hopp and B. Muschielok. timate that there will be over 15 Tera- in these key bands. When other filters
byte of raw data per year. This raw data are used, the calibration plan foresees
3.4 Control Software volume contains roughly 5 Terabyte of a cross calibration of these filters ver-
calibration data and 10 Terabyte of raw sus these key bands.
All instrument functions (filter ex- science data. Data processing will then The basic technique to overcome
change, shutter, detector readout, as produce another 10 Terabyte of re- any gaps or artefacts in the CCD pixels
well as monitoring the instrument state) duced science data and may create, is to take more exposures of the same
are controlled in software. The pro- with about 100,000 astronomical ob- field with slightly shifted field centre and
gramming environment is defined and jects per OmegaCAM field, enormous to co-add the images off-line. We dis-
provided by ESO through the releases catalogues. To efficiently handle this tinguish the following observing modes:
of the VLT Common Software which data volume the data acquisition, cali- • Dither has offsets matching the
has to be used as the basis for design brations and the pipeline reductions are maximum gap between CCDs, ∼ 400
and development. The partitioning of strictly procedurized, a key aim being to pixels (5.6 mm). It will be operated with
the OmegaCAM Instrument Software
(OmegaCAM INS) into software sub-
systems also follows the VLT stan-
dards. Nevertheless there were several
challenges peculiar to OmegaCAM.
The Autoguiding Software and Image
Analysis modules normally belong to
the Telescope Control Software. In the
case of OmegaCAM it was necessary Figure 5:
to move these functionalities to the INS Theoretical
because during normal operations the throughput
VST guiding arm will not be used, as it curves from
slightly vignets the science array. A new SAGEM for the
SDSS filter set,
software algorithm was developed to
and measured
extract optical aberration coefficients quantum
from the out-of-focus images recorded efficiency of one
on the Image Analysis CCDs. On the of the
detector software side, particular atten- OmegaCAM
tion had to be paid to the coordination CCDs.
17
N (with 5 as the default value) pointings The observing modes and strategies Acknowledgements
on the sky. Although this will nearly cov- are fully integrated with the data reduc-
er all the gaps in the focal plane and tion software being developed by the The consortium was formed in 1998
maximizes the sky coverage, the con- OmegaCAM consortium. We distin- in response to an announcement of op-
text map of such data is complex. An guish between a calibration pipeline portunity from ESO, and comprises in-
advantage is that it will be relatively producing and qualifying calibration stitutes in the Netherlands (NOVA, in
easy to couple the photometry among files, and an image pipeline that applies particular the Kapteyn Institute Gronin-
the individual CCDs. the calibration files to raw data and gen and Leiden Observatory), Ger-
• Jitter has offsets matching the transforms them into astrometrically many (in particular University Observa-
smallest gaps in CCDs ∼ 5 pixels. This and photometrically calibrated images. tories of Munich, Göttingen and Bonn)
mode optimizes the homogeneity of the ESO users will be provided with the and Italy (INAF, in particular Padua and
context map and will be used during ob- output of the image pipeline, run in Naples observatories). The ESO Op-
servations for which the wide gaps are Garching, on the data contained in a tical Detector Team provides the detec-
not critical, but which, for instance, re- single OB. The nominal photometric tor system at cost to the consortium.
quire a well-mapped smoothly varying accuracy of this pipeline will be ± 0.05 OmegaCAM is headed by PI K. Kuijken
PSF. mag, exceptionally ± 0.01 mag. The (Groningen and Leiden University) and
• Stare allows re-observing one fixed nominal accuracy for the astrometry is co-PI’s R. Bender (Munich USM/MPE)
pointing position multiple times. It is the ± 0.1 arcsec rms over the entire field of and Cappellaro (INAF Naples/Padua),
main workhorse for monitoring the in- view. and project management is done by B.
strument and allows detection of optical As part of the contract, the Muschielok and R. Häfner (USM).
transients. OmegaCAM consortium will deliver OmegaCAM is funded by grants from
• SSO is the mode for observing software modules that ESO will inte- the Dutch Organization for Research in
Solar System Objects, which requires grate into the image pipeline. In addi- Astronomy (NOVA), the German Fed-
non-sidereal tracking. tion, a project has been set up among eral Ministry of Education, Science, Re-
For all these modes dedicated ob- European wide-field imaging groups to search and Technology (grants 05
serving templates are being developed. provide a ‘wide-field imaging survey AV9MG1/7, AV9WM2/5, 05 AV2MGA/6
An observing strategy employs one system’ that will combine pipeline pro- and 05 AV2WM1/2), and the Italian
or a combination of the basic observing cessing of image data with archiving Consorzio Nazionale per l’Astronomia
modes. It also defines a number of ad- and data mining tools. Further details e l’Astrofisica (CNAA) and Istituto Na-
ditional instructions for scheduling of can be found on http://www.astro- zionale di Astrofisica (INAF), in addition
the observations. We distinguish the wise.org, and in Valentijn & Kuijken to manpower and materials provided by
following strategies: (2002). the partner institutes.
• Standard which consists of a single The development of the analysis
observation (observation block) software is being done by a team based References
• Deep which does deep integrations, in Groningen and Leiden, led by E.
possibly taken at selected atmospheric Valentijn. Arnaboldi, M., Capaccioli, M., Mancini, D.,
Rafanelli, P., Scaranella, R., Sedmak, G.
conditions over several nights and Vettolani, G. P., 1998. The Mes-
• Freq which frequently visits (moni- 5. Current Status senger 93, 30.
tors) the same field on time scales Baruffolo, A., Bortolussi, A., De Pizzol, L.
ranging from minutes to months and The OmegaCAM project is now well 2002, Proc. SPIE 4848, in press.
Nicklas, H., Harke, R., Wellem, W., Reif, K.
has overriding priority on the telescope into in the manufacturing phase. Most 2002, Proc. SPIE 4836-34, in press.
schedule of the CCDs have been delivered and Reif, K., Klink, G., Müller, Ph. and Posch-
• Mosaic maps areas of the sky larg- tested; most of the mechanics exist and mann, H. 2002 in Scientific Detectors for
er than 1 degree, which is essentially are ready to be integrated; instrument Astronomy, Beletic, Amico eds., Astro-
an item for the scheduling, as the control and data analysis software is physics and Space Sciences Library
(Kluwer: Dordrecht), in press.
pipeline has to produce uniform quality being coded. Extensive tests in Europe Valentijn, E.A. & Kuijken, K. 2002 in Toward
data anyway. The combination of vari- are foreseen for the second half of an International Virtual Observatory,
ous field centres into one image is not 2003, and the camera should see first Quinn, P., ed., ESO Astrophysics Sym-
considered a standard pipeline task. light early in 2004. Exciting times! posia Series (Springer-Verlag), in press.
18
combiner IONIC for VINCI, the accept-
ance of the science instrument MIDI
in Heidelberg in September (currently
being integrated at Paranal), and three
new contracts to extend the VLTI infra-
structure: Delay Lines 4, 5 and 6 were
ordered at Fokker Space in Leiden, and
the installation of the required rail sys-
tems in the Delay Line Tunnel started in
August. The contract for the fourth
Auxiliary Telescope was signed with
AMOS in Liège at the beginning of
September (see [7]). The first PRIMA
contract for the delivery of the fringe
sensor unit was signed with ALENIA
Spazio, Torino, in July, and the second
contract for the star separator unit will
follow soon. The PRIMA laser metrolo-
gy system will be an in-house develop-
ment. These three subsystems will form
the first phase of PRIMA.
IONIC was provided by the Observa-
toire de Grenoble. IONIC is a beam com-
biner for VINCI that can be used instead
of VINCI’s fibre beam combiner MONA.
In IONIC, the beams are guided through
two silicon fibres into an integrated op-
tics (IO) component where the beam
combination takes place. At the exit of
the IO component, fibers are attached
guiding the light to the infrared detector.
Since the fibre connectors are identical
to those used in MONA, the optical in-
tegration was just a matter of decon- Figure 1: The reconstructed, two-dimensional interferometric point spread function (PSF) of
necting and connecting a few fibres. the star Achernar observed in the K-band. The width of the central fringe is 3 by 15 milli-arc-
Unlike the K-band beam combiner sec indicating the angular resolution limit of the VLTI for the baseline distribution shown in
MONA, IONIC, having silicon fibres and Figure 2. On the largest scale, the image is enveloped by the Airy disk of a single 8-m Unit
substrates, works in the H-band. The Telescope. Its first minimum at 57 milli-arcsec off the centre can be clearly seen. The image
VLTI had first fringes with IONIC on provides a dramatic illustration of the 20-fold increase in resolution of the VLTI over a single
July 18, producing an interferometric 8.2-m telescope.
transfer function above 80% with an
accuracy of 1% which is as good as with the Unit Telescopes. The intensity livers two function values, at (u, v) and
MONA (see [8]). distribution was reconstructed from the at (-u, -v).
Last but not least there is also measurements with individual baselines The balloons representing the meas-
progress with the adaptive optics sys- by Fourier transforming the distribution ured visibilities in the uv-plane are
tems for the VLTI. The Paranalization of of values displayed in Figure 2. Every small disks although a point would be
the tip-tilt sensors STRAP on two Unit balloon in that figure represents the re- expected when observing with an indi-
Telescopes is finished, and MACAO, sult of a visibility measurement with one vidual baseline. The reason is that the
the high-order adaptive optics system baseline vector. It is important to prop- telescope is not a pinhole but an 8-m
(see [9]), saw first light in the laborato- erly refer to baselines as baseline vec- aperture allowing for a small variation
ry in Garching in August, closing the tors in order to understand why the dis- of baselines around the ’centre-to-cen-
loop on an artificial star. tribution of the balloons is considerably tre’ baseline. Inside VINCI, optical fi-
different from the distribution of Unit bres are used as spatial filters. The size
2. Four Eyes Are Better Telescopes at Paranal (see Fig. 3). The of the optical fibres is exactly an Airy
visibility function depends on the differ- disk of the telescope, i.e. the field of
Why should one attempt to use more ence of telescope coordinates (project- view is one Airy disk. This spatial filter-
than two telescopes if the interferomet- ed perpendicular to the viewing direc- ing is equivalent to performing an aver-
ric instrument can only combine two tion of the star). For instance, two tele- age over the visibilities inside the disk
beams? First of all, this exercise scopes separated by 100 m along the represented by the balloons.
demonstrated the ability to provide mul- north-south direction have exactly the For imaging it is important to note that
tiple beams in the VLTI laboratory with- same baseline vector as two other tele- one measures as many independent
out losing time for mirror realignment or scopes a few kilometres away that are image points inside the Airy disk as one
for other reconfigurations. In 2003, AM- also separated by 100 m along north- has baselines, i.e. as one has degrees
BER, the near infrared science instru- south. of freedom. Fortunately, the number of
ment, will see first fringes. Then we will The baseline vectors are displayed in baselines is not only determined by the
have an instrument combining up to the ‘uv-plane’, where the measured number of telescopes but also by the
three telescopes simultaneously. visibility forms the function value at the duration of the observation. Due to the
In addition, multiple baselines are the coordinate given by the length and sidereal motion, the baseline projected
key to efficient interferometric imaging. the orientation of the baseline vector. onto the viewing direction changes all
The result of the observations of the Since the visibilities are derived from the time in length and orientation de-
star Achernar in September, shown in the intensity, a real and positive quanti- pending on the object coordinates.
Figure 1, gives an impression of the an- ty, the visibility function is point sym- Some of the baselines that are dis-
gular resolution capability of the VLTI metric; one visibility measurement de- played in Figure 2 have a slight peanut
19
N
shape due to the observing time of a 8-m telescope shows up in the point four Auxiliary Telescopes. With six
few hours. Several individual meas- spread function in Figure 1. The first Delay Lines, a common focus of six tel-
urements were taken during this time. dark ring of the Airy disk with a radius escopes can be used at any given mo-
Therefore, the number of individual of 57 milli-arcsec in the K-band is read- ment. To fully exploit the VLTI infra-
points in the image will be considerably ily apparent. This illustrates the gain in structure, the first second-generation
larger than six when observing with all angular resolution with the VLTI com- instrument should clearly be a six-way
four Unit Telescopes. pared to an individual Unit Telescope. beam combiner, measuring 15 base-
The present observations of Acher- However, it is still a long way to go be- lines at once and 28 during a night,
nar give an impression of the point fore the VLTI will do imaging of arbitrary without reconfiguring the telescopes.
spread function of the VLTI because objects since this requires measuring Amongst all these integration and
Achernar with an angular diameter of 2 the phase of the visibility function in ad- commissioning activities one must not
milli-arcsec is smaller than the resolu- dition to its modulus i.e. its contrast. forget that the success of an observa-
tion limit of 3 milli-arcsec of the VLTI in AMBER in 2003 will be a first step, be- tory depends solely on the scientific
the K-band. Due to the moderate num- ing able to use three telescopes and output. MIDI will make a start next year,
ber of baselines, the point spread func- performing a phase reconstruction via and 2004 will be largely devoted to sci-
tion in Figure 1 shows a mix of a mod- the closure phase technique. Eventu- ence operations with the VLTI.
erate number of individual fringe pat- ally, from 2005 onwards, PRIMA will al-
terns. A larger number of visibility low measuring the phase directly per Acknowledgement
measurements would make the individ- baseline and, in addition, observe very Over the last 20 months, the number
ual fringe patterns disappear, and only faint objects. The scientific objectives of of people both at ESO and in the com-
the central fringe would remain as the the PRIMA facility are described in [10]. munity contributing to the VLTI has be-
‘point’ in the point spread function. come too large to name here. I would
The particular distribution of baselines 3. Outlook like to thank not only those with the
in Figure 2 is rather extended in the The coming year will be extremely highest visibility but also the large num-
north-east/south-west direction (the busy, with integrations of new subsys- ber of contributors forming the base of
baseline of ANTU-YEPUN) and rather tems about every two months. FINITO, the success by ensuring that every little
narrow in the perpendicular direction the fringe sensor unit, will be first, fol- piece of this complex machine works
because the baseline MELIPAL-YEPUN lowed by two MACAO systems, by the every night.
is missing. Consequently, the Fourier first two Auxiliary Telescopes, by AM-
transform of this distribution is rather BER and by the carriages with cat’s References
narrow in the north-east/south-west di- eyes of the three new Delay Lines. This
rection and rather wide in the perpendi- is a big burden for the integration teams [1] Kervella, P., et al. 2002, in Exotic Stars
cular direction. The terms narrow and since the VLTI is not the only ‘tele- as Challenges to Evolution, IAU Sym-
posium 187, in press.
wide refer to the width of the central scope’ at Paranal. [2] Wittkowski, M. and Hummel, Ch. 2002, in
fringe. The central fringe in Figure 1 has In 2004, the remaining two MACAO Interferometry for Optical Astronomy II,
a width of about 3 milli-arcsec in the di- systems and the third and fourth ed. Traub, W., Proc. SPIE 4838, in press.
rection of the longest baseline and of Auxiliary Telescopes will arrive at [3] Richichi, A. and Wittkowski, M. 2002, in
about 15 milli-arcsec in the perpendicu- Paranal so that the VLTI will be able to The VLTI: Challenges for the Future, eds.
lar direction. Due to the processing combine the light from eight different Garcia P.J.V., Glindemann A., Henning
technique, the Airy disk of an individual telescopes, four Unit Telescopes and T., Malbet F., JENAM Workshop, in press.
20
[4] Segransan, D., et al. 2003, A&A Letters, J. V., Glindemann A., Henning T., Malbet [9] Arsenault, R., et al. 2002, in Adaptive
in press. F., JENAM Workshop, in press. Optical System Technologies II, eds.
[5] Paresce, F., et al. 2002a, in Inter- [7] Koehler, B., et al. 2002, The Messenger, Bonaccini, D., Wizinowich, P., Proc.
ferometry for Optical Astronomy II, ed. this volume. SPIE 4839, in press.
Traub, W., Proc. SPIE 4838, in press. [8] Kern, P., et al. 2002, in Interferometry for [10] Paresce, F., et al. 2002b, in Interferom-
[6] Percheron, I., et al. 2002, in The VLTI: Optical Astronomy II, ed. Traub, W., etry for Optical Astronomy II, ed. Traub,
Challenges for the Future, eds. Garcia P. Proc. SPIE 4838, in press. W., Proc. SPIE 4838, in press.
1. Introduction This article recalls the rationale at the m versus a maximum of 130 m be-
origin of the ATS development, pro- tween UTs. This is needed to reach the
In June 1998, ESO signed a contract vides a description of the design and fi- ultimate angular resolution of the VLTI
with the company AMOS (Belgium) for nally reports on the performance as that scales as λ/B (i.e. 0.6 milli-arcsec
the supply of the Auxiliary Telescope measured in Europe so far. in the visible and 2 milli-arcsec in the K
System (ATS) for the VLTI. The original band).
scope covered two movable Auxiliary 2. Why Does the VLTI Need • It enables full-time use of the VLTI
Telescopes (ATs), as shown in Figure 1, Auxiliary Telescopes? facilities, since the ATs are entirely ded-
as well as their associated site equip- icated to interferometric observations.
ment including rails and interface de- The VLTI is primarily intended to This is an important factor for the VLTI
vices for each observing station. An combine coherently the four VLT 8-m scientific productivity (and for the amor-
amendment was signed in September Unit Telescopes (UTs). Obviously, the tization of its development cost!).
1999 for the supply of a third AT and, ultimate VLTI sensitivity will indeed be • It is required by the Narrow
last September, a fourth AT was or- obtained when combining the UTs. Angle Astrometry mode of PRIMA that
dered. However, the array
The contract with AMOS is based on of 1.8-m diameter
top-level performance requirements ATs is a key ele-
and includes the design, manufacturing ment for the tech-
and testing of the complete system in- nical and scientific
cluding all mirrors and cells, complete capability of the in-
mechanical structure, drives, encoders, terferometer. The
small mechanisms and low-level elec- main reasons are
tronics. The main sub-contractors of listed below.
AMOS are FISBA (Switzerland) for the • It provides the
coudé mirrors, PHASE (Italy) for the best imaging capa-
motors and CSEM (Switzerland) for the bility of VLTI by
M2 hexapod mechanism. complementing the
ESO is in charge of the design and array of UTs. It
development of the telescope control gives access to 30
hardware and software, as well as of telescope stations
the two star sensors located at the increasing there-
coudé focus. fore the number of
After integration of the ESO control accessible base-
hardware and software, the telescope line vectors, a fun-
is fully tested in Europe at AMOS with damental parame-
the possibility of sky observation pro- ter for high-fidelity
vided by an outside observing station image reconstruc-
included in the dedicated test facility. tion.
After this, ESO transports the ATs to • It gives access
Paranal, re-integrates them in the so- to the longest VLTI
called Mirror Maintenance Building baseline of B = 200
(MMB) and finally commissions them
on the sky.
The project entered into manufactur- Figure 1: The first
ing phase in mid-1999 and is now Auxiliary Telescope
reaching the end of a very extensive for the VLTI during
testing and verification phase in Europe final testing at
on AT#1 before its shipment to Paranal AMOS in Liège
next year. (Belgium).
21
by using the so-called ‘boot-strapping’
technique where the fringes are tracked
on the relatively short intermediate
baselines while the low-visibility fringes
are integrated on the extreme baseline.
In addition, the fourth AT will permit to
dedicate two ATs for the PRIMA as-
trometry mode while two others are be-
ing used by MIDI or AMBER. This may
prove a fundamental advantage during
the challenging commissioning phase
of PRIMA by alleviating the possible
conflicts with the regular VLTI science
observation.
22
▲ Figure 3: The 1.8-m primary mirror in its cell during integration into
the telescope at AMOS. The mirror is lightweighted to reduce its
mass by a factor about 2. A 54-points wiffle-tree axial support was
selected to comply with the stringent image quality requirements.
infrastructure such as the total mass 3.3 Telescope structure on a complex strut structure with ap-
and the design volume are also driving propriate welded box/plates interfaces
factors for the overall AT design. An overview of the telescope design to the altitude shaft, Serrurier structure
can be seen in Figure 2. The main driv- and M1 cell. On each side of the cen-
3.2 Optical design ers were the maximization of the first trepiece, a semi-circular plate holds the
eigenfrequency, the minimization of the magnets for the altitude motors and
The AT optical design is similar to M1-M2 de-space and de-centre (typi- tachometers and a brake disk. For the
that of the UT and includes 11 mirrors. cally < 10 µm) under gravity, wind load support of M3, a spider was preferred
The first three (M1-M3) constitute a and thermal load as well as the mini- over a more classical tower solution be-
Ritchey-Chrétien telescope with a 1.8- mization of the wind torque. This had to cause it provides better static and dy-
m diameter primary mirror. A fast f/1.4 be achieved while complying with the namic results while easing the re-align-
primary mirror was selected to provide constraints of mass, design volume, sta- ment after an M1 removal for re-coating.
a compact tube design and limit wind tion interface and earthquake survival. The tube is connected to the Fork on
buffeting of particular concern for inter- The Tube structure is a classical both sides by double row angular con-
ferometric application where nanome- Serrurier strut made of hollow tubes tact ball bearings providing high stiff-
ter-level stability is required in the connecting to a top-ring made out of full ness, low run-out and low friction. The
Optical Path Length (OPL). One conse- rods for weight balance. The M2 spi- Fork is a welded box structure with in-
quence of that particular design is the ders were optimized in order to limit the ternal ribs highly optimized for stiffness
high sensitivity of the telescope to the design volume, to reduce the torque and mass. At its lower part, the Fork
secondary mirror (M2) defocus & de- due to wind load on the top-ring and to connects to the Azimuth bearing, a key
centre which places high demand on increase axial stiffness in view of OPL element to the overall telescope stiff-
the stiffness of the tube structure. The stability. A particular feature of the tube ness. The Azimuth bearing is a three-
Nasmyth focus of the Ritchey-Chrétien design is the solution adopted, after row roller bearing with very high radial,
is unused for normal operation. It is re- several trade-offs, for the thermal com- axial and tilt stiffness but, at the same
imaged by the 5-mirror coudé train (M4- pensation rendered necessary by the time, very low friction torque necessary
M8) to form a coudé focus located un- extreme sensitivity to M2 defocus. The for the pointing/tracking accuracy.
derground. In addition, the coudé train compensation is provided by a proper Below the Azimuth bearing, the
creates an image of the telescope pupil selection of material for the Serrurier Ground Interface Structure (GIS) pro-
on M6 where a fast tip-tilt mirror (and (steel), top-ring (steel) and M2 spider vides the connection to the ground
possibly later an adaptive optics mirror) (invar). When the temperature increas- through the pseudo-kinematic anchor-
is located. Just above the coudé focus, es, the tube expands and the top-ring ing and clamping devices. This part of
a dichroic mirror (M9) transmits the vis- diameter increases. This later creates a the telescope was also intensely opti-
ible to a CCD camera and an APD mod- stress in the M2 spiders that results in mized by FEM due to its importance in
ule (STRAP) for acquisition and fast a downward motion of M2 compensat- the first telescope eigenfrequency.
guiding. The infrared light reflected by ing the tube and M1 cell expansion. Inside the station pit the Relay Optics
M9 is collimated by the Relay Optics Several trade-offs were made on the Structure (ROS), a double-skin welded
(M10-M11) to form an 18mm parallel centrepiece that is a main contributor to box cylindrical structure, houses the
beam sent towards the VLTI delay stiffness and wind torque. The latter Relay Optics and the coudé sensors.
lines. constraint led to the final design based When relocating an AT on a new sta-
23
Figure 5: The first telescope on its test bench inside the dedicated Figure 6: The Transporter in its relocation configuration. The cylin-
hall at AMOS. In addition to tests at component level such as optical der in the foreground houses the Relay Optics and the coudé
quality of each mirror, friction of the axis bearings, etc., the first tel- sensors (CCD and STRAP). The light beam exits the telescope by
escope undergoes a thorough acceptance test programme including one of the two black openings. During relocation, the cylinder is low-
modal tests of the complete structure, dynamic response of the azi- ered underground inside the station pit before the telescope is
muth and altitude axis, OPL stability measurement, verification of struc- placed on top.
tural deformation under gravity, pointing repeatability check and others.
tion, this structure is lowered into the The axis position information is given counter-weights. In addition, three radi-
station hole on soft pads, the telescope by high accuracy Heidenhain optical al fixed points are located inside the M1
is placed on top and a centring and encoders. The high interpolation factor central hole.
clamping mechanism is activated be- leads to the very high angular resolu- The M1 cell consists mainly of a
tween the structure and the Ground tion of 9 milli-arcsec on Altitude and 3 welded steel structure with a double
Interface Structure. The position re- milli-arcsec on Azimuth. skin for the bottom and lateral parts to
peatability is such that local alignment improve the cell stiffness. The structure
is not required after relocation. This 3.5 Primary mirror unit is made of two parts to facilitate the mir-
holds also for the repeatability of the ror integration and maintenance. The
telescope anchoring system. The only The M1 unit is shown on Figure 3. It cell’s lower part interfaces with the axi-
telescope alignment necessary after re- consists of the primary mirror, its cell al support while the upper part inter-
location is performed remotely from the with axial and lateral supports and fi- faces with the lateral support. The first
Control Room. nally the interface with the telescope eigenfrequency of the M1 unit is 20 Hz
tube. and the mirror degradation in its cell un-
3.4 Drives, tachometer and The primary mirror is made out of a der gravity is less than 11 nm RMS sur-
encoders 200-mm thick blank of Zerodur material face.
from SCHOTT (Germany). Due to
The motors selected for both Altitude mass constraints, the mirror is light-
and Azimuth axes are direct drive, weighted by drilling hexagonal pockets 3.6 Secondary mirror unit
brushless torque motor, similar in con- on its rear face as can be seen on
cept to those installed in the UT. Two Figure 3. It is polished to a hyperbolic Due to the selected fast f/1.4 primary
curved linear motor pads are used dia- shape. mirror, the secondary M2 mirror is a
metrically opposed for Azimuth and one The mirror axial support is the result rather small mirror 140-mm in diameter.
pad is attached to each side of the fork of an intensive optimization by Finite However, its highly aspherical and fast
for Altitude. Each motor is equipped Element Modelling in order to minimize (f/1.3) shape makes it a difficult mirror
with internal position sensors used in the effect of gravity deflection. The se- to manufacture. It is also made out of a
particular for motor commutation. lected solution compatible with the blank of Zerodur, 25 mm thick.
Each axis is equipped with an AC-DC lightweight geometry and the low-level The M2 is mounted on a hexapod
analogue tacho-generator using the error budget is a 54-axial-point support support (see Fig. 4) providing defocus,
same magnet strip and similar pads as mounted on a three-stage wiffle-tree de-centring and tip-tilt motions with
the motor. The power amplifier is based system using flexible pivots. sub-micron and arcsecond level resolu-
on fast DSP technology and provides For the lateral support, the retained tion for remote alignment. Its design is
motor current control, overspeed limita- optimal solution consists of 16 astatic primarily driven by the very limited
tion and tacho demodulation. levers using low-friction bearing and space available (in the shadow of M2)
24
and the requirement to limit cross cou- Modules that are fitted in-between the The auxiliary power module consists
pling between focus and tilt motion. two cylinders. The four drive wheel of a set of batteries feeding a UPS to
blocks, attached to the outer cylinder, provide uninterrupted power to the con-
3.7 Enclosure include the electrically driven wheel trol electronics as well as an electrogen
and hydraulically driven lifting, rotation group connected to the UPS to provide
The basic enclosure concept is to and blocking mechanisms. The tele- electric power during relocation when
provide good natural ventilation during scope lifting device consists of 8 lifting the telescope is unplugged from its sta-
the night to limit “dome and mirror see- actuators linked to cardan shafts which tion socket.
ing” and therefore to leave the part of are driven by two electric motors. The The hydraulic group provides the
the telescope above the altitude axis Relay Optics handling device uses two required oil pressure and flow for the
exposed to the outside wind (see Fig. electric linear actuators linked by a various hydraulic mechanisms. Such
1). Following an analysis of the wind floating shaft. mechanisms are used for the Trans-
load, thermal requirements and mass The other functions of the Trans- porter wheels 90° rotation, the Trans-
and volume constraints, the selected porter are to provide air conditioning, porter lifting, the clamping of Trans-
design consists of composite spherical cooling liquid, auxiliary power as wells porter to ground, Telescope to ground
shells that open up with two hydraulic as hydraulic and pneumatic power for and Relay Optics Structure to Telescope,
cylinders attached to each of the two the various mechanisms. It also pro- as well as for the Enclosure opening/
upper shells. When closed, the two up- vides the space for the control elec- closing.
per shells are clamped together by tronics cabinets. Finally, a pneumatic system is used
means of 6 electrical actuators and in- The air conditioning system provides to inflate/deflate the Enclosure seals
flatable seals are inflated in-between temperature control during the day and to activate the Telescope axes
each shell. The normal status of the en- based on three temperature sensors lo- brakes.
closure is fully open or fully closed. cated on the primary mirror, plus six
However, it is possible to close totally or other sensors distributed on the tele- 3.9 Station equipment and rails
partially any of the two sides in order to scope structure. The mirror tempera-
act as a wind shield in case of high wind ture is controlled to ±0.5°C of ambient The station equipment consists of a
coming from S-E or N-W, the two dom- air with the capacity to cool the mirror set of Telescope interfaces, Transporter
inant wind directions. by 2°C in about 3 hours. The cooling interfaces and a station lid. The tele-
power is injected with filtered, re-circu- scope interface is made of a plate
3.8 Transporter and Service lated air through a circular fabric duct whose upper face is machined to pro-
Modules with constant flow and high induction to vide a perfect supporting plane for the
ensure temperature homogeneity. The telescope anchoring counterparts.
The main function of the Transporter control is done through a PLC with an These counterparts are male elements
is to provide the handling means to re- optimal set point computed by the ESO consisting of a clamping finger placed
locate the telescope without the need AT Control System based on outside inside a cone. They come into the cor-
for any additional external devices. A temperature measurement and fore- responding conical part of the tele-
typical relocation sequence can be cast for the next night. scope clamping device to form a pseu-
summarized as follows: (i) unclamp the The liquid cooling module is provid- do- kinematic interface. The station Lid
Relay Optics Structure from the tele- ing cooling power during observation is a composite hexagonal cover fitted
scope, (ii) unclamp the Telescope, (iii) lift for temperature control of electronic with a seal. It ensures good thermal in-
the Telescope, (iv) unclamp the Trans- cabinets and heat-dissipating equip- sulation, air tightness and resists to a
porter, (v) lower down the Transporter ment such as motors, coudé sensors 500-kg central load. It can be handled
wheel on the rails and lift the Trans- and small mechanisms. The remaining by two persons.
porter in driving position, (vi) drive the heat dissipated in the vicinity of the op- The rails are standard rolled steel
Transporter beside the station, (vii) at- tical beam is maintained below 25W profile with end cut at 45° to ensure
tach the Relay Optics to its handling de- and surface temperatures are kept at smooth load transfer at the junctions. At
vice integrated on the Transporter and ambient ±1.5 °C. In order to avoid the a crossing, a specific design enables to
lift it, and (viii) drive to the next station. vibration of the chiller during observa- transfer the load from the central rolling
When a change of direction is needed tion, the cooling
(at one of the rail crossing), the Trans- energy is stored in-
porter is stopped at the crossing, low- side a tank of heat
ered and clamped on special interface latent nodules. The
plates. The wheels are lifted up and ro- cooling capacity is
tated by 90° before being lowered on 1 kW during 12 h
the perpendicular rails. Unclamping and is re-generat-
and lifting the Transporter enables to ed during the day
drive to the next crossing. At the desti- by running the
nation station, the reverse operations chiller.
are performed.
The Transporter (see Fig. 2) consists
of a frame structure made of hollow alu-
minium rectangular profiles. It has been
optimized with respect to mass mini- Figure 7: The first
mization (aluminium), space con- Transporter sees the
straints and earthquake survival during sun light for the first
observation and relocation (with tele- time at AMOS (Liège,
scope load). An inner cylindrical struc- Belgium) during its
functional tests last
ture supports the enclosure and the
October. A dummy
Telescope lifting devices. An outer station will permit
cylinder structure, covered with alu- testing of the tele-
minium panels and equipped with ac- scope on the sky be-
cess doors, provides protection of the fore shipment to
electronics cabinets and the Service Paranal.
25
Figure 8: Installation of the AT interface devices inside each of the 30 observing stations at Paranal. A high-precision alignment method us-
ing theodolite and specific tools is used to reach a positioning accuracy below a millimetre and an arcminute.
surface of the wheel to its two lateral However, two significant new ap- cellent achievement for the first mirror
flanges. At each track end, an end-stop proaches have been applied on the AT. of this size and short focal ratio manu-
is installed. On the Hardware side, a fully digital factured by AMOS.
control scheme was selected in place Currently, the first AT is in its last test-
3.10 Control system of the former analogue velocity con- ing phase at AMOS. The telescope, ful-
trollers for the telescope axes. This was ly integrated with optics, is anchored on
The design & implementation of the one of the first applications of what has a dedicated station (see Fig. 5) where
control system is shared between become now a new VLT standard: the all the major performance tests have
AMOS and ESO. AMOS is responsible generic Tool for Advance Control (TAC). been completed, as described in the
for the Transporter control including the Its main advantage is the possibility to next section. Meanwhile, the Trans-
complete relocation sequence and the configure and optimize the control algo- porter has also been fully integrated
Service Modules while ESO is in rithm in a very flexible manner even at and the second telescope – partially in-
charge of the complete Telescope Con- runtime. TAC is now used for several tegrated – has been placed inside in
trol including control of the axes, the M2 applications in the VLT/VLTI (OPD order to proceed with an extensive
support, the field stabilization loop as Controller, FINITO). On the Software functional test phase during which all
well as the small telescope mecha- side, the Unified Software Develop- motions required for the relocation
nisms such as M10. ment Process with the Unified Modeling have been exercised and tuned (see
On the AMOS side, two Programma- Language (UML) has been introduced Fig. 7). One of the main remaining ac-
ble Logical Controllers (PLC’s) perform at the beginning of the AT project. tivities is the testing of the relocation
the control tasks. The first one (Sauter) performance, including the time to relo-
is dedicated to the Air Conditioning and 4. Current Status cate (less than 3 hours) and the re-
Liquid Cooling Module while the sec- peatability of the telescope positioning
ond one (Siemens) controls the other As far as optics are concerned, the full (± 100 µm and ± 10 arcsec). The latter
Transporter functions including those set of 11 mirrors is complete for each of was already successfully tested on a
needed during relocation. For the latter, the first three ATs except the primary breadboard during the design phase.
the operation is done locally from a mirrors for AT#2 and AT#3 that are re- The other important remaining test con-
portable control panel (see Figure 7). spectively under final figuring and ma- cerns the performance of the thermal
Both communicate with the ESO Local chining. The blank for AT#4 has been control provided by the air conditioning
Control Units (LCU’s) for exchange of ordered and light-weighting will start during the day and by the liquid cooling
status information and commands before the end of the year. Each mirror module during night observation. In
when needed (e.g. switch on/off air has been successively polished within January, the formal technical accept-
conditioning, set reference tempera- its very tight allocated specification de- ance in Europe of the first AT will take
ture, open/close enclosure, etc.) rived from the glob-
On the side of ESO, the control sys- al telescope image
tem adheres to the Hardware and quality given in
Software standards adopted for the Table 1. As an ex-
VLT. It is based on a distributed archi- ample, the phase
tecture of workstations (Unix) connect- map of the first pri-
ed to the LCU’s (VxWorks) via a LAN mary mirror for AT#1
(Ethernet). For each AT, the control is shown on Figure 9.
electronics located on-board the The achieved wave-
Transporter is connected through the front error amounts
station plug and the LAN optical fibres to less than 24 nm
to the VLT or VLTI Control Room from RMS. This is an ex-
where all control is done during regular
operation (i.e. except for relocation).
Figure 9: The 1.8-m
The design of the Software makes primary mirror of AT#1
extensive re-use of the Telescope after polishing. The
Control Software (TCS) implemented wavefront error is be-
on the UT with a few modifications re- low 24 nanometers
quired by the difference in hardware. RMS
26
Figure 11: Overview of measured non-repeatable pointing error in
arcsec on the sky for various angular motions of the altitude and az-
imuth axes. The circles represent the requirement specifications for
the various angular ranges.
place. Subsequently, the system will be the signals from the tachometer or the bination of laser interferometer and ac-
handed over to ESO for a 3-month test position encoder is a critical character- celerometer measurements. The M1-
period in Europe where the ESO istic for an accurate, high-bandwidth M2 OPL variation was measured by
Control System will be tested and fine- control of the telescope axes (under the means of two high-sensitivity ac-
tuned with a possibility for sky observa- responsibility of ESO). For this reason, celerometers while the OPL variation
tions. The first AT will then be packed the dynamic responses of both altitude from M2 to M11 was measured by
and sent to Paranal where re-integra- and azimuth axes have been specified means of an HP Doppler interferome-
tion will start mid-2003. and measured. Figure 10 shows the re- ter. The results of the OPL test were
The second and third telescopes are sult of the measurement to which a combined together with the other con-
currently in the assembly process. Most simplified dynamic model of the tele- tributors to the OPL stability as comput-
of the sub-contracted equipment for the scope could be fitted in order to verify ed by analysis such as the wind load
telescopes such as the main drives, M2 that the control bandwidth and stability and the micro-seismic disturbances.
hexapods, bearings are integrated. The criteria will be met when ESO plugs in The consolidated budget appears to be
Transporter for AT#2 is being assem- its telescope control system. The results compliant with the specifications.
bled while the one for AT#3 is still under are well within the specifications of 8 Hz
manufacturing. bandwidth for the velocity loop and 2 5.3 Verification of non-repeatable
At Paranal, the preparatory activities Hz bandwidth for the position loop. pointing error
are taking up momentum in order to be
ready to receive the first AT. The tele- 5.2 Verification of the The pointing accuracy is particularly
scope interface devices have been in- Optical Path Length stability important for a telescope dedicated to
stalled and carefully aligned in each of interferometry. Indeed, the standard in-
the 30 stations (see Fig. 8) and the rail One of the most unusual and difficult terferometric observing strategy con-
installation is about to start. requirements for the AT is the stability sists in switching rapidly (every 1 to 3
of the Optical Path Length (OPL) inside minutes) between the scientific object
5. First Results from Telescope the telescope that is required by inter- and a calibrator of known diameter that
Performance Tests ferometric observations. The OPL vari- is used to calibrate the time varying in-
ation was a driving parameter of the tel- strumental and atmospheric visibility
In order to verify system-level per- escope mechanical design where opti- losses. Poor pointing repeatability re-
formance requirements before the final mization processes were necessary to quiring full acquisition sequences can
commissioning on site, an extensive fulfil the specifications. The identified therefore seriously affect the efficiency
test program has been carried on AT#1 contributors to the OPL variation are the of the observation. A test was dedicat-
at AMOS. Some of the most important vibrations resulting from the dynamic ed to measure the non-repeatable part
results are reported below. wind load, the normal micro-seismic ac- of the AT pointing accuracy for five
tivity and the Transporter and telescope pointing directions distributed over the
5.1 Verification of the telescope active sources during operation. In or- operational range. A collimator and five
axis dynamic response der to verify the contribution of the tele- folding mirrors were installed above the
scope internal vibrations to the total telescope. The collimated light, sent via
The dynamic transfer function be- OPL stability budget, a dedicated OPL one folding mirror, is focused at the
tween the motor torque reference and test set-up was designed using a com- coudé focus on a CCD. The image po-
27
sition is recorded before and after the 6. Conclusion related performance tests are cur-
telescope has been moved back and rently in progress. Early next year, the
forth by a given angle (from 10 arcmin- This article provides an overview first AT will be ready for a short ESO
utes, up to 60 or 360 degrees) in alti- of the current development of the test period in Liège to tune the tele-
tude and azimuth directions. Due to the Auxiliary Telescopes for the VLTI. The scope control system developed in
very small pointing errors to be meas- main performance tests of the first tele- parallel at ESO. The year 2003 will
ured (0.1 to 1 arcseconds), particular scope are completed. They have see the installation at Paranal of the
attention had to be paid to the thermal shown that the system is on track to first two ATs. This will considerably
environment and to the processing meet the severe and specific require- boost the scientific productivity of VLTI
method in order to eliminate thermal ments originating from its dedication that will, at that time, be equipped
drifts of the measurement set-up. to interferometry, in particular a high dy- with its two first-generation instru-
Figure 11 gives an overview of the ex- namic performance. The functional ments: MIDI and AMBER. AT#3 is
cellent results obtained, basically limit- tests related to the relocation of the tel- expected to be ready for scientific
ed by the residual measurement noise escope have been successfully com- observation in mid-2004 and AT#4 in
of 0.13 arcsec. pleted on the first Transporter and the early 2005.
We live in interesting times. Last year 2.9% on Melipal, and 3.1% on Yepun. unwieldly named instrument, is widely
was our first with all four 8-m tele- Observing time losses due to adverse recognized as the most powerful adap-
scopes in operation. We had only one weather conditions represented 12% of tive optics facility instrument in the
instrument per telescope but we were the science time, somewhat more than world. Already fabulous results have
kept busy with UVES and ISAAC work- in previous years. We archived 57,810 appeared in press releases and more
ing round the clock and commissioning frames on ISAAC, 34,329 frames on are to come. The spectacular results on
and installations during bright time UVES, 22,853 frames on FORS1 and the Galactic centre have demonstrated
when FORS1 and FORS2 were taking 37,847 frames on FORS2. the excellent capabilities of the instru-
a breather. Now at the end of 2002 we NAOS/CONICA has been brought ment. UT4 now has two instruments
look at Paranal and a very different pic- into operation in a major effort by the and both dark and bright time are fully
ture appears in front of us. consortia that built it, ESO instrumenta- exploited for science.
We continued to operate with low tion division and Paranal. All parties VIMOS appeared, somewhat over-
technical time losses. In the period from pulled out all the stops to start science weight and with all the challenges of
April to October 2002 the down time operations on the 1st of October 2002. what in effect is four imaging, multiob-
was 3.6% on Antu, 4.4% on Kueyen, NACO, as we now call the otherwise ject spectrographs mounted on a single
frame. Starting with two arms, it then
had a leg added to it and another two
arms making the whole thing quite a
sight to be admired on the Nasmyth B
platform of UT3. All of this under the
careful supervision of Paranal engi-
neering and extensive testing and dis-
cussions to ensure that nothing unto-
ward occurred as we pushed the rota-
tors beyond their specified and tested
range. Again with a lot of work from the
consortium and the support of the in-
strumentation division and the ever
present Sandro D’Odorico, we have
now reached the stage of Paranaliza-
tion of the instrument and we have a
fixed date with our customers on the 1st
of April 2003. Already the PI (Olivier Le
Fevre) has claimed success with the
execution of the entire CFR Survey ex-
ecuted in a single exposure.
In the mean time FLAMES arrived in
parts and has slowly grown to occupy
the totality of the Nasmyth A platform of
UT2. First OzPoz, the fibre positioner
and then GIRAFFE the spectrograph to
take the 130 spectra at medium resolu-
tion simultaneously. OzPoz did puzzle
us for a while. Mostly it puzzled the con-
sortium that built it. After a few inter-
ventions on the instrument to make
sure it fit to the telescope properly and
Figure 1: The VST enclosure taking shape at the edge of the Paranal platform. to make sure it could cope with the real
28
operational temperatures, it worked
well enough to give us confidence to of-
fer the instrument to the community
also on April 1st 2003. Paranalization is
in full swing also for FLAMES and the
visits from Australia to fix it are rare.
FORS1 moved back to UT1 to make
dark time available to VIMOS. Now
UT1, UT2 and UT4 have dark and
bright time instruments attached and ei-
ther offered or just about to be. UT3
with VIMOS is eagerly awaiting VISIR
next year.
On UT4 we added a laser clean
room, kindly provided by telescope divi-
sion in Garching. This is a whole new
platform under the Nasmyth B focus of
UT4 with a fancy room to house the
laser for the artificial guide star to be in-
stalled at the VLT in 2003. We mounted
a small telescope on the back on the
secondary of UT4 to check for flexures
and entertained ourselves and Martin
Cullum trying to work out where the
thing was pointing. In any case the flex-
ures are low and we await the laser and Figure 2: The primary of UT3 coming out of the coating chamber with a reflectivity of 92%
the fibre to carry the light plus the and a micro-roughness of 7 Ångströms. The best coating ever at Paranal.
launch telescope during 2003.
This would be enough for most ob- the time of writing these pieces are writing we are removing the primary
servatories to be going on. However, starting to be placed on top of the VST and tertiary from UT1 and plan in the
Paranal is more than a home for four 8- concrete pier for the enclosure of our next few days to have the fourth 8-m
m telescopes. The interferometer con- next telescope. The enclosure is ex- coating this year. Moving these mirrors
tinues to grow in capabilities and im- pected to be finished in early 2003 and around and going through the cleaning
proves its operations. Last year we the telescope is to be installed soon af- and coating process, along with all the
started the installation and first runs ter. A few weeks before the enclosure preventative maintenance that is per-
with the tip-tilt boxes at the coudé foci. erection begun we received the boxes formed on the domes and telescopes
Stabilized beams were fed from UT1 containing the VST 2.6-m primary and while the glass is out, is a major engi-
and UT3 into VINCI and excellent secondary mirrors. Unfortunately the neering effort organized to the minute
fringes were obtained. In 2002 the sys- primary mirror was destroyed during and synchronized with great precision.
tems were stabilized operationally with transport. Our colleagues at the Osser- The observatory is now undertaking the
very few nights of sky time. The other vatorio di Capodimonte in Naples who maintenance of the passive supports
two coudé trains (UT2 and UT4) were are building the telescope very quickly for the cell in an effort that is supported
installed aligned and tested. The read- ordered a replacement mirror and if all by La Silla who are providing the work-
er should be aware that 11 mirrors ap- goes well we expect no significant de- shop space.
pear in the normal coudé train before lay to the project. In January of 2002 we all moved into
the beam is sent on the interferometer We welcomed the VISTA site super- the new Paranal residence. The long
delay line tunnel. In September we ob- visor to Paranal and we prepare for the stay in the containers is finally over and
tained fringes with all four telescopes start of the works to accommodate the we now have a pleasant and comfort-
marking a major milestone in the devel- 4-m infrared wide field telescope. able environment in which to live. Work
opment of the interferometric capabili- The smallest telescope on Paranal on Paranal is by necessity associated
ties of Paranal. was also commissioned. Called Mascot with long absences from family and
In November, MIDI, our first fully (ask J.G. Cuby) it is an all-sky camera friends and extreme weather condi-
fledged interferometric instrument ar- based on simple commercially avail- tions, in particular very low humidity.
rived on site and is being integrated as able components and with some in The new residence attempts, and we
this article is being written. First light house ingenuity it makes its images believe succeeds, to reduce the hard-
should arrive soon. AMBER, FINITO available in real time on the UT con- ships associated with work on site.
and full coudé adaptive optics on the soles. The operators
UTs for VLTI are all in the plan for 2003. still go out to have a
Three additional delay lines are be- look.
ing installed into the tunnels and vari- Coating mirrors on
able curvature mirrors added to the Paranal is a never
system to better control the position of ending story. How-
the pupil. The VLTI complex is reaching ever, in 2002 we have
a maturity of operation at Paranal that reached a peak of ac-
is comparable to that of the UTs. We tivity. At the time of
eagerly await the arrival of the first aux-
iliary telescope next year and the
preparations on Paranal are ongoing
Figure 3: All-sky image
with the alignment of the docking sta- from Mascot Camera
tions and the installation of the tracks. which monitors the
In September of 2002, a set of boxes cloud cover on Paranal
arrived on Paranal containing suspi- continuously during ob-
ciously Paranal blue metal pieces. At servations.
29
Some of the containers are being re- casino serves approximately 100,000 the ALMA project. With this move a
tained as we continue to build on meals every year and we have 40,000 small shake up of the observatory or-
Paranal (e.g. VST and VISTA) and of- overnight stays per year. Paranal is not ganization was necessary and it has
ten have more people working on site connected to the Chilean electricity grid been a pleasure to welcome Frank
than are expected in a steady state of and has to generate its own power Ruseler from the Santiago office of
operation. The recent MIDI commis- (up to three megawatts). Water is of ESO to Paranal where he has taken
sioning required for the reactivation of course a necessary resource and we over the logistics department.
some containers for ESO and consor- can store up to 1 million litres in our The year ahead of us promises to be
tium staff. Not all were happy with the tanks. Keeping the observatory sup- as exciting as this past one. VISIR,
move back to the not-so-good-old days. plied requires a water tanker truck to ar- VST, OmegaCAM, MACAOs for the
The new Visitor’s Centre was con- rive on site every 8 hours, every day of UTs, FINITO, AMBER, Auxiliary tele-
cluded and is in use for the weekend the year. scopes and the laser guide star all in-
visitor programme, and as a starting In 2002, Joerg Eschwey, who more tend to arrive on Paranal. Times will
point for VIP visitors. than any other individual has personi- continue to be interesting on Paranal
The scale of the Paranal operation is fied the challenge and success of cre- for the foreseeable future. Exciting
often difficult to appreciate. Ourselves ating a whole observatory and its infra- new facilities and capabilities continue
on the mountain find it all fairly normal. structure in the middle of the Atacama to be added to this astronomy wonder-
To put it into context the ESO Paranal desert, has moved on from the VLT to land.
30
trol rooms of the NTT, 3.6-m and 2.2-m
telescopes in a common area, allowing
us to operate the telescopes more effi-
ciently.
The RITZ was built with the comfort
of its inhabitants in mind. Toward the
west, it has many large windows over-
looking the valley and ocean, with a
fantastic view of the sunset (and poten-
tial green-flashes). On the other hand,
its walls are almost completely blind to-
ward the north-east since this is the di-
rection from which the wind blows most
of the time. All of the control computers
are located in separate rooms behind
the consoles, allowing us to keep the
computers cold and the control room
temperate. These computer rooms also
provide additional thermal insulation for
the main working area, and keep it free
from the noise of the many worksta-
tions. The floor plan of the main control
room, as well as the materials chosen
for the walls and ceiling are ergonomi-
cally optimized, ensuring that sounds
are damped and glare from the win-
dows is avoided. efficiency of the mosaic is significantly result in an improvement of the spec-
Since October 18, the NTT has been higher than that of the former Tektronix trograph’s performance by a factor of 2.
operated fully remotely from the RITZ. CCD and the larger size of the CCDs Keep an eye on the FEROS pages for
We hope to move the 2.2-m control leads to a wider field of view. In addi- up-to-date information.
room this year, and the 3.6-m early next tion, the pixel size (0.166″) gives a bet- TIMMI2, the mid-IR spectro-imager
year. ter sampling in good seeing images mounted at the 3.6-m, is currently un-
and permits narrower slits to be used. dergoing a major upgrade: the control
News from the Instruments and However, the most striking feature of electronics and software are being en-
Telescopes the new detector is its read-out time: tirely replaced by an IRACE controller
thanks to the FIERA controller, the 4k and VLT-software. This will improve the
From the start of Period 70 (start of chips are read in only 18 seconds, with performance of the instrument, but
October), and as reported in previous very low read-out noise. More informa- more notably, it will make its operation
Messenger articles, the two 1.5-m tele- tion can be found on the updated EMMI and maintenance much more straight-
scopes on La Silla are no longer offered web pages. forward. Also, for the observer, the
to the ESO community. The ESO 1.52- FEROS, the high-resolution, fibre- main difference will be that TIMMI2 will
m will be operated until the end of 2002, fed spectrograph that used to be be operated from the standard P2PP-
with 100% of the time allocated to mounted on the ESO 1.52-m telescope, BOB interfaces.
Brazilian observers, and the Danish has been transferred to the 2.2-m, Finally, ADONIS, the adaptive optics
1.54-m will continue to be used by the where it is now installed in parallel with system at the 3.6-m, was retired at the
Danish community only. the Wide Field Imager. The move took end of August after almost 10 years in
A few months ago, the red CCD of place in early October and the instru- service. The field of high resolution im-
EMMI was upgraded to a mosaic of 2 × ment is currently being commissioned aging is now carried by NAOS-CONICA
1 MIT/LL 2k × 4k chips. The quantum in its new home. We expect that this will on the VLT.
Two unusual views of La Silla. The left-hand picture shows La Silla immediately after the snow storm of 28 August 2002. Some weeks lat-
er the desert around La Silla was for a short time transformed into a sea of flowers. Photos by Peter Sinclaire.
31
REPORTS FROM OBSERVERS
Extra-Solar Planets
N.C. SANTOS, M. MAYOR, D. QUELOZ, S. UDRY
Observatoire de Genève, Sauverny, Switzerland
Introduction densate, these “planetesimals” are Orion stellar nursery, showing that
thought to grow in a few million years. disks are indeed very common around
The widely accepted picture of stellar When such a “planetesimal” achieves young solar-type stars. This supports
formation tells us that a planetary sys- enough mass (about 10 times the mass the idea that extra-solar planets should
tem is a simple by-product of the stellar of the Earth), its gravitational pull is suf- be common. However, such systems
formation process. When a cloud of gas ficiently strong for it to start accreting have escaped detection until very re-
and dust contracts to form a star, con- gas in a runaway process that gives cently.
servation of angular momentum in- origin to a giant gaseous planet similar In fact, it was not until 1995, following
duces the formation of a flat disk to the outer planets in our own Solar the discovery of the planet orbiting the
around the central newborn “sun”. By a System. Later on, in the inner part of solar-type star 51Peg May95, that the
process still not fully understood, this the disk, where temperatures are too search for extra-solar planets had its
disk is believed to be the stage for the high and volatiles cannot condensate, first success1. This long wait was main-
planetary formation. According to the silicate particles are gathered to form ly due to the difficulty of detecting such
traditional paradigm, dust particles and the telluric planets like our Earth. bodies. Planets are cold objects, and
ice grains in the disk are gathered to In the past decade, images taken by their visible spectrum results basically
form the first planetary seeds (e.g. the NASA/ESA Hubble Space Tele- from reflected light of the parent star. As
Pollack et al. 1996). In the “outer” re- scope (HST) have revealed a multitude a result, the planet/stellar luminosity ra-
gions of the disk, where ices can con- of such proto-planetary disks in the tio is of the order of 10–9. Seen from a
distance of a few parsec, a planet is no
more than a small “undetectable”
speckle embedded in the diffraction
and/or aberration of the stellar image.
The detection of exoplanets has thus
been based, up to now, upon “indirect”
methods. In particular, all the planetary
discoveries were only possible due to
the development of high-precision radi-
al-velocity techniques. These methods,
that measure the motion of a star along
the line of sight (by measuring the
Doppler shift of spectral lines), have
now achieved precisions of the order of
a few m s–1 (∆λ/λ ∼ 10–8). Such a high
precision is indeed needed to find a
planet: for example, Jupiter induces a
periodic perturbation with an amplitude
of only 13 m s–1 on the Sun!
In this article we will review the cur-
rent status of planetary searches, pre-
senting the major challenges that we
are facing at this moment. We will then
discuss how new and future generation
instruments and missions will help to
answer the most important questions.
We will concentrate mostly on the re-
sults we can expect from future radial-
velocity campaigns with state-of-the-art
instruments like HARPS on the ESO
3.6-m telescope (see article by Pepe et
al. in this issue).
A Diversity of Planets
Figure 1: Schematic orbital configurations for some of the newly found extra-solar planets in
three different scales. In the upper-left panel we represent the orbits of the shorter-period Today, about 100 extra-solar plane-
companions. The Sun (yellow circle) is drawn to scale. This plot illustrates well both the prox- tary systems have been unveiled
imity of these planets to their host stars, and the complete lack of planetary companions or-
biting closer than a certain distance (see text for more details). The upper-right and lower pan- 1
Before this discovery, only planets around a
els illustrate the situation concerning longer orbital period planets. In these two plots, the or- pulsar had been detected (Wolszczan & Frail 1992);
bits of the Earth and Jupiter are also drawn for comparison (with the usual symbols). These however, these are probably second-generation
three panels clearly illustrate the huge variety of orbital parameters presented by the extra- planets. In this article we will concentrate on plan-
solar planets. ets around solar-type stars.
32
around stars other than our Sun2. Figure 2: Distribu-
These discoveries, which include ∼ 10 tion of minimum
masses for the cur-
multi-planetary systems, have brought rently discovered
to light the existence of planets with a low-mass compan-
huge variety of characteristics, opening ions to solar-type
unexpected questions about the stars. Although the
processes of giant planetary formation. radial-velocity
The diversity of the discovered extra- method has a
solar planets is well illustrated in higher sensitivity to
Figure 1. Unexpectedly, they don’t higher-mass com-
have much in common with the giant panions, the
observed distribu-
planets in our own Solar System. tion rises very
Contrarily to these latter, the “new” steeply towards the
worlds present an enormous and unex- low-mass domain.
pected variety of masses and orbital From this mass up
parameters (astronomers were basical- to the stellar
ly expecting to find “Jupiters” orbiting at regime, only a few
∼ 5 A.U. or more from their host stars in objects were
quasi-circular trajectories). The majori- detected; this
ty of the discovered planets were not region is usually
denominated the
even supposed to exist according to the “Brown-Dwarf
traditional paradigm of giant planetary desert”. This gap in
formation (e.g. Pollack et al. 1996). the mass distribu-
Their masses vary from sub-Saturn to tion of low-mass
several times the mass of Jupiter. companions to so-
Some have orbits with semi-major axes lar-type stars supports the view that the physical mechanisms involved in the formation of
smaller than the distance from Mercury these two populations (planets vs. stars) are very different.
to the Sun, and except for the closest
companions, they generally follow ec-
centric trajectories, contrarily to the In order to explain the newly found the disk and consequently stops the mi-
case of the giant-planets in the Solar systems, several mechanisms have gration), photo-evaporation, tidal inter-
System. thus been proposed. Current results action with the host star, or Roche-lobe
These findings have put into question show that in situ formation is very un- overflow of the young inflated giant
the former planetary formation para- likely, and we need to invoke inward mi- planet (processes resulting in an in-
digm. However, the relatively large num- gration, either due to gravitational inter- crease of the orbital radius of the
ber of discovered planets is already action with the disk (Goldreich & planet, thus opposing the migration
permitting us to undertake the first sta- Tremaine 1980) or with other compan- tendency)3 .
tistical studies of the properties of the ions to explain the observed orbital pe- In any case, even if these mecha-
exoplanets, as well as of their host stars. riods. In other words, the observed nisms are able to explain how the short-
This is bringing new constraints to the close-in planets could simply have er period planets have stopped migra-
models of planet formation and evolu- been formed far from their host star, tion, they do not explain how the longer
tion. Let us then see in more detail what and then migrated inwards. period ones (like Jupiter itself) did not
kind of problems and information these But the migration mechanisms, that migrate to distances closer to their host
new discoveries have brought. have broken long-lasting ideas of “sta- stars. The key to this might have to do
bility” of the planetary systems, have with a combination of parameters, like
The Period Distribution some problems to solve. According to the disk masses, lifetimes, viscosities,
the models, the timescales of planetary and initial planetary masses and/or
One of the most interesting problems migration in a disk are particularly number of bodies formed, that will af-
that appeared after the first planets short. This means that more than wor- fect the final orbital configuration of a
were discovered has to do with the rying about how planets migrate after or planet. Some of these parameters are not
proximity to their host stars. The first during their formation, we need to un- well known (maybe the planetary for-
planet discovered (around 51Peg – derstand how migration can be stopped mation itself is controlling them), some-
Mayor & Queloz 1995) is exactly the (and/or slowed down)! thing that complicates the discussion.
first such example: it orbits its star once One particularly interesting clue
every 4.2 days, corresponding to an or- comes from the observation that there The Mass Distribution
bital radius of only 0.05 A.U. This is is a clear pile-up of planetary compan-
much less than the distance from ions with periods around 3 days, ac- Very important information is brought
Mercury to the Sun. companied by a complete absence of to us by the analysis of the mass spec-
The problem that was raised with the any system with a period shorter than trum of the planetary companions. In
finding of these 51Peg-like planets this value. This result, which is in com- particular, a plot showing the mass dis-
(usually called “hot-Jupiters”) resides in plete contrast with the period distribu- tribution of companions to solar-type
the fact that at such close distances the tion for stellar companions (we can find stars (as shown in Figure 2), shows a
temperatures are too high for ices to double stars with periods much shorter clear discontinuity for the mass regime
condensate, and there does not seem than 3 days), means that somehow the between about 30 and 50 times the
to exist enough available material to process involved in the planetary mi- mass of Jupiter: there are basically no
form a Jupiter-mass planet. It is thus gration makes the planet “stop” at a dis- companions found to date having those
very difficult to imagine that such tance corresponding to this orbital peri- masses. This result is even more strik-
worlds could be formed so close to the od. To explain this fact, several ideas ing if we note that the radial-velocity
central stellar furnace. have been presented. These invoke dif-
ferent mechanisms like e.g. a magne- 3Planets migrating more than this approximate
2For a complete and updated list of the known
tospheric central cavity of the accretion limit might “simply” also evaporate/transfer materi-
exoplanets, see e.g. table at http://obswww.unige.ch/ disk (once the planet gets into this cav- al to the host star, and thus disappear or become
exoplanets. ity it will no longer strongly interact with too low-mass to be detected.
33
Figure 3: Distribution sis of the orbital eccentricities of the
of exoplanet masses. planetary-mass companions. Accord-
The histogram repre- ing to the traditional paradigm of plane-
sents the observed tary formation, a planet (formed in a
minimum mass distri-
bution, while the red
disk) should keep a relatively circular
line represents the (low eccentricity) trajectory. Current
statistical true plane- models shown indeed that the interac-
tary mass distribution tion (and migration) of a low mass com-
resulting from a de- panion within a gas disk has the effect
convolution of the un- of damping its eccentricity (Goldreich &
known orbital inclina- Tremaine 1980). However, opposite to
tions. The distribution expectations, if we look at the eccen-
reaches “zero” at a tricities of the planetary companions we
mass of about 10
times the mass of
can see that they are spread through
Jupiter, which probably values that go from nearly zero to more
corresponds to the than 0.9 for the planet orbiting the star
upper limit for the mass HD80606!
of a giant planet. The In Figure 4 we plot the eccentricity as
nature of the objects a function of orbital period for the plan-
with masses between etary companions to solar-type stars,
10 and ∼17 times the as well as for the stellar companions.
mass of Jupiter is still First of all, it is important to describe the
an open question. As
in Jorissen et al.
general tendencies observed in the
(2001). plot. The low eccentricity observed for
short-period binaries is the result of a
well known effect: the proximity to their
technique is more sensitive to massive panions (M > 2 MJup) orbiting in short primary stars induces tidal interactions
companions than to their lower mass period (lower than ∼ 40-days) trajecto- that have the effect of damping the ec-
counterparts. ries. Similarly, there seems to be a lack centricity. Since the tidal effect de-
This gap, usually called the “brown of long-period and very low-mass creases very fast with distance, above
dwarf desert”, separates the low mass planets. a given orbital period (about 10 days for
“planetary” companions from their high These results are further helping us dwarf star binaries), tidal circularization
mass “stellar” counterparts, and is to understand the mechanisms of plan- is no longer effective, and all compan-
probably telling us something very im- et migration, since they are compatible ions having periods longer than a given
portant about the physical processes with the current ideas about planetary value simply keep their “initial” orbital
involved in the formation of these two orbital evolution (either due to an inter- eccentricity.
populations: stars, even the low mass action with the disk or with other com- While both distributions show the sig-
ones, are thought to be formed as the panions), that suggest that the higher nature of tidal effects on the eccentrici-
result of the gravitational collapse and the mass of a planet, the more slowly ty, a first glimpse also tells us that there
fragmentation of a cloud of gas and (and less) it will migrate. is no clear difference between the two
dust. On the other hand, a planet forms groups of points: stars and planets
in a circumstellar accretion disk. The Eccentricity have a similar distribution in this dia-
More information is provided if we gram. This poses the problem of un-
analyse the shape of the distribution for One of the most enigmatic results derstanding how planetary companions
the planetary mass regime (Fig. 3). found to date has to do with the analy- formed in a disk can have the same ec-
This distribution is observed to de-
crease smoothly with increasing mass, Figure 4: Eccentricity
reaching “zero” at about 10 Jupiter vs. orbital period (in
masses (Jorissen et al. 2001). This lim- logarithmic scale) for
it is clearly not related to the Deuterium- the discovered extra-
burning mass limit of ∼ 13 MJup, some- solar planets (red
times considered as the limiting mass pentagons), stellar
binaries (filled dots),
for a planet (this latter value is in fact an
and for the giant
arbitrary limit used as a possible “defi- planets in our Solar
nition”, but it is not related to the plane- System (green stars).
tary formation physics). As it was re- The earth is repre-
cently shown by several authors (e.g. sented by the usual
Jorissen et al. 2001), this result is not symbol. Although
an artefact of projection effects (the un- some long-period ex-
known orbital inclination implies that we oplanets exist having
can only derive a minimum mass for the low values for the
companion from the radial-velocity eccentricity, most of
measurements), but a real upper limit the systems present
for the mass of the planetary compan- much higher eccen-
ions discovered so far. tricities than those
observed for the
Solar System giant
The Period-Mass Relation planets. Possible ex-
planations invoke
Recent results have also unveiled mechanisms capable
some interesting correlations between of pumping the
the planetary mass and its orbital eccentricity, like
period. In fact, there seems to exist a gravitational interactions between planets in a multi-planetary system, or with a distant
paucity of high-mass planetary com- stellar companion.
34
centricity distribution as their stellar
mass counterparts. And how can this
be fit into the picture of a planet forming
(and migrating) in a disk?
The explanation for these facts may
be other processes capable of exciting
the eccentricity of the planetary orbits.
These include the interaction between
planets in a multiple system or between
the planet and a disk of planetesimals,
the simultaneous migration of various
planets in a disk, or the influence of a
distant stellar companion. All or at least
some of these physical processes
might play an important role in defining
the “final” orbital configuration.
Although still not clear, a close inspec- Figure 5: Left: Metallicity ([Fe/H]) distributions for stars with planets (yellow histogram) com-
pared with the same distribution for field dwarfs in the solar neighbourhood (open red his-
tion of Figure 4 permits us to find a few togram). In this panel, both distributions are normalized by the total number of points. The
differences between the eccentricities [Fe/H] scale is logarithmic, and the Sun has by definition [Fe/H] = 0. Most planet hosts are
of the stellar and planetary compan- more metal-rich than our Sun; Right: the percentage of stars that have been found to harbour
ions. In particular, for periods in the a planet for each metallicity bin, plotted as a function of the metallicity. This plot clearly shows
range of 10 to 30 days (clearly outside that the probability of finding a giant planet increases with the metallicity of the star. As in
the circularization period by tidal inter- Santos et al. (2001).
action with the star), there are a few
planet hosts having very low eccentric- formation of the currently discovered surface is still deeply mixed with the hot
ity, while no stellar binaries are present extra-solar planetary systems. stellar interior (this is not the case when
in this region. The same and even Boss (1997) has suggested that be- the star reaches its “adulthood”). 6Li is
stronger trend is seen for longer peri- sides the core accretion scenario (see thus not supposed to exist in stars like
ods, suggesting the presence of a introduction), giant planets might also HD 82943, and the simplest and most
group of planetary companions with or- be formed as a result of disk instability convincing way to explain its presence
bital characteristics more similar to processes: by the formation and con- is to consider that planet(s), or at least
those of the planets in the Solar Sys- densation of clumps of gas and dust in planetary material, have fallen into HD
tem (with low eccentricity and long the protoplanetary disk. This process 82943 sometime during its lifetime.
period). On the other hand, for the is, however, not very dependent on the The most recent and detailed analy-
very short period systems, we can see metallicity. In other words, if the disk-in- sis seems to clearly confirm the pres-
some planetary companions with ec- stability models were the most impor- ence of this isotope. The question is then
centricities higher than those found for tant mechanism involved in the forma- turned to know whether this case is iso-
“stars”. This features may be telling us tion of giant planets, we should not ex- lated or if it represents a frequent out-
that different formation and evolution pect to see a strong dependence on the come of the planetary formation process.
processes took place: for example, the rate of planet detection as a function of How much can this process increase
former group may be seen as a sign for the metallicity. The huge dependence the observed metallicity of the planet
formation in a disk, and the latter one observed is thus probably a sign that hosts? Current results suggest that at
as an evidence of the influence of a the core accretion scenario is the im- least the degree of stellar “pollution” is
longer period companion on the eccen- portant mechanism involved in the for- not incredibly high (Santos et al. 2001).
tricity. mation of giant planets.
It is, however, important to stress that Black Sheep
Clues from the Planet Hosts: it is not precisely known how the metal-
the Stellar Metallicity licity is influencing the planetary forma- When measuring the spectrum of a
tion and/or evolution; for example, the star we are obtaining the integrated
Up to now we have been reviewing masses of the disks themselves, which light of the whole stellar disk, and gath-
the results and conclusions obtained di- can be crucial to determine the efficien- ering photons coming from different
rectly from the study of the orbital prop- cy of planetary formation, are not points in the stellar surface. Each indi-
erties and masses of the discovered known observationally with enough vidual point has its own spectrum, with
planets. But another fact that is helping precision. a different Doppler shift that is a func-
astronomers understand the mecha- tion of the velocity field in that specific
nisms of planetary formation has to do A Case of Stellar “Cannibalism” region of the stellar photosphere. As a
with the planet host stars themselves. consequence, any phenomenon capa-
Indeed, they were found to be particu- Recent observations also suggest ble of changing the velocity field of a
larly metal-rich, i.e. they have, on aver- that planets might be engulfed by their given region in the stellar surface will
age, a higher metal content than the parent stars, whether as the result of change the global spectrum Doppler
stars without detected planetary com- orbital migration, or e.g. of gravitational shift, and consequently the measured
panions (see Santos et al. 2001 for the interactions with other planets or stellar radial-velocity.
most recent results) – see Figure 5. companions. Probably the clearest evi- This result has an important impact
A possible and likely interpretation of dence of such an event comes from the when dealing with radial-velocity meas-
this may be that the higher the metallic- detection of the lithium isotope 6Li in the urements: the radial-velocity technique
ity of the cloud that gives origin to the atmosphere of the planet-host star HD is not sensitive only to the motion of a
star/planetary system (and thus the 82943 (Israelian et al. 2001). This frag- star around the centre of mass of a
dust content of the disk), the faster a ile isotope is easily destroyed (at only star/planet system, but also to eventual
planetesimal can grow, and the higher 1.6 million degrees, through (p,α) reac- variations in the structure of the stellar
the probability that a giant planet is tions) during the early evolutionary surface.
formed before the proto-planetary disk stages of star formation. At this stage, In fact, phenomena such as stellar
dissipates. In other words, the metallic- the proto-star is completely convective, pulsation, inhomogeneous convection
ity seems to be playing a key role in the and the relatively cool material at the or the presence of dark spots (e.g. Saar
35
Figure 6: Radial-ve- The increase in long-term precision,
locity measurements and the continuation of the current
and best 2-planet high-precision radial-velocity pro-
Keplerian solution for grammes, will also give us the opportu-
HD 82943. The radi-
al-velocity measure-
nity of finding more and more long-pe-
ments show that this riod planets. Current models predict
star has a system of that the planetary formation and evolu-
two resonant plan- tion processes should produce more
ets, orbiting it with long-period planets than their short pe-
periods of ∼ 220 and riod counterparts. In fact, the current
440 days, respec- surveys are just starting to discover
tively. This same star exo-planets with periods comparable to
was found to have the ones found for the Solar System gi-
6
Li in its atmosphere,
a signature of possi-
ants. This is an essential goal in order
ble planet engulf- to improve the statistics of these ob-
ment (Israelian et al. jects, and to check if the Solar System
2001). is anomalous or common.
Of course, other problems will arise
and become more important as a con-
& Donahue 1997) are expected to in- inition of a planet itself is currently un- sequence of the dramatic increase in
duce radial velocity variations. Further- der debate. precision. In particular, the intrinsic stel-
more, contamination from the spectrum To solve this problem, there are high lar radial-velocity “jitter” produced by
of other stellar companions can also in- expectations from new or soon-to-be e.g. chromospheric activity related phe-
duce spurious radial-velocity signals. available instruments (see the paper on nomena (e.g. Saar & Donahue 1997)
These cases can prevent us from find- the HARPS spectrometer in this issue). might impose serious (and still un-
ing planets (if the perturbation is larger The incredible precision gain achieved known) limits on the final precision, and
than the orbital radial-velocity varia- by these radial-velocity “machines” will on the consequent ability to find very
tion), but perhaps more importantly, be crucial in this aspect. But what are low-mass (and long-period) systems.
they might give us false candidates if exactly the important lines of study to This is particularly true for the young-
they produce a periodic signal (e.g. a follow? And what kind of answers can est stars. Some effort should thus
rotating stellar spot). The radial-veloci- we expect in the next few years? be put into the development of diag-
ty technique is thus most efficient for nostics capable of confirming the plan-
low chromospheric activity single “old” Increasing the Statistics etary nature of the radial-velocity sig-
dwarfs. nal. Furthermore, we might even imag-
A few good examples of spurious pe- As we have seen above, the ob- ine that the spurious radial-velocity
riodic radial-velocity variations can be served correlations between the orbital variations caused by activity might be
found in the literature. The first to be parameters of the newly found planets modelled and corrected, leaving only
described was the periodic signal ob- are giving astronomers a completely the real gravitational effects on the
served for the dwarf HD 166435, that different view on the formation and evo- signal.
was shown to be due to the presence of lution of planetary systems. We no
a dark spot rather than to the gravita- longer have the Sun as the only exam- Planets Around M Dwarfs
tional influence of a planetary compan- ple, and today we have to deal with the
ion. peculiar characteristics of the “new” ex- Although more than 100 exoplanet
At the current (and increasing) level tra-solar planets: a huge variety of peri- candidates are now in the lists, only two
of precision obtained using the radial- ods, eccentricities, masses. planetary companions around M dwarfs
velocity techniques, such kinds of ex- To clear up current uncertainties we have been detected (both in the solely
amples might become more and more need more data. Only a large and sta- system Gl 876). The very low number
common. It is thus very important to de- tistically uniform set of data may enable of M-dwarf planetary companions can
velop ways to disentangle e.g. activity- us to clarify the current situation. In this in fact be largely explained by observa-
related phenomena from real planetary sense, the hundreds of new planets ex- tional biases: the very low mass dwarfs
candidates. Such methods might be pected to be discovered in the next few are faint and it is difficult to obtain ac-
based on the study of the shape of the years will have a very important out- curate radial velocities for them.
spectral lines (usually called the “bisec- come. However, to constrain the various
tors”), the photometric variability of the scenarios of planetary-system forma-
star, and/or of chromospheric activity Lower and Lower Masses tion and evolution, it is now crucial to
indexes. This is very important for proj- obtain better statistics for planets
ects like HARPS, for which the barrier One clear result of the increase in around the most numerous stars in our
of the 1 ms–1 will be achieved. precision of the current and future radi- galaxy. M dwarfs compose 80% of the
al-velocity surveys is the ability to find main sequence stars. How many of
Guidelines for the Future lower-mass planets. them have planets? How these planets
As we have seen above, the mass differ from those orbiting the more
The study of extra-solar planetary function for planetary companions massive G dwarfs is totally unknown.
systems is just beginning. After only 7 around solar-type stars is rising toward These questions await the future capa-
years, we can say that at least 5% of the low-mass regime. But given that the bilities of instruments like HARPS.
solar-type dwarfs have giant planetary radial-velocity technique is more sensi-
companions in relatively short period tive to the more massive companions, The Chemical Link
orbits (≤ 3 years). However, and as we the lower mass bins in the plot of
have seen in the previous sections, Figure 2 are definitely not well repre- Planet host stars seem to be, on av-
many interesting but troubling problems sented. How then does this function be- erage, particularly metal-rich. This in-
still await a solution. In fact, the newly- have for the lower-mass regime? What teresting result probably reflects the im-
found planets have clearly disturbed is the minimum mass of a giant planet? portance the quantity of available rocky
the long-standing theories of giant plan- How does this depend on the orbital pe- material in the disk has on the forma-
etary formation and evolution. The def- riod? tion of giant planets.
36
As discussed above, this link might star HD 209458, a dwarf that was pre- On the other hand, the strong inter-
hold the key to understanding how gi- viously discovered to host a very-short action between planets in such sys-
ant planets are formed. The two com- period planetary companion (Charbon- tems will be reflected as an observable
peting theories (core accretion and disk neau et al. 2000). evolution of their orbital parameters. A
instability) should have different sensi- This detection, and the subsequent dynamical analysis of this will give us
tivities to the metallicity. If most planets related studies, have had an enormous the opportunity to obtain information on
were formed by the latter of these two impact for the understanding of these the masses and relative orbital inclina-
processes we should not expect any systems. For example, it was possible tions for the companions.
special metallicity sensitivity. to estimate the mean density of the
Although the metallicity trend is planet, and to prove that it is orbiting in Planets in Binaries
clearly seen (Santos et al. 2001), there the same direction and plane as the
are nevertheless a few planet hosts that star’s equator. To date, several planets have been
are particularly metal poor (see e.g. case In the near future we can expect that discovered in known multiple stellar
of HD 6434). How can this be explained? other such events might be brought to systems. Moreover, a fraction of stars
One elegant way of solving this puzzle light. In particular, many hundreds of known to host planets exhibit a drift in
is to look for the frequency of planets photometric transit candidates are ex- the γ-velocity indicating the presence of
around metal-poor halo dwarfs. If, as pected from space missions like an additional distant companion.
expected, no planetary mass objects COROT, Kepler or Eddington. How- These observations show that de-
are found around such a sample of ever, based only on photometry we spite the gravitational perturbation of
stars, then the disk instability model is cannot determine whether the ob- the stellar companion, planets may form
clearly put into question. However, if served transiting body is a planet or and survive around stars in multiple sys-
some are found, giant planets might be simply a low-mass star (since the effect tems. The properties of such planets
formed by different processes. is of similar magnitude because of the hold important clues on the mecha-
There are also some traces of stellar large degeneracy of the radius of these nisms of planetary formation. For ex-
“pollution” among the planet host stars objects). The follow-up of the photo- ample, according to the standard core
(Israelian et al. 2001). This opens the metric observations by radial-velocity accretion model of planetary formation,
question of understanding how much surveys is thus essential and will permit a giant planet is formed by the accre-
planetary material might fall into the us to obtain the real mass of the planet. tion of gas around a ∼ 10 earth mass
convective envelope as a consequence With such data we can hope to derive core of rocky material. This is supposed
of the planetary formation process it- empirical relations between variables to take place at distances comparable
self. How much can this change the such as the planet’s mean density and to the Jupiter–Sun separation (∼ 5 A.U.).
metallicity of the star? If important, this its distance from the star, its mass, and Opposing this model, Boss (1997) has
could have consequences even for the stellar metal abundance. In this proposed that giant gaseous planets
studies of the chemical evolution of the sense it is important to say that the very might also be formed from the conden-
galaxy. high precision of HARPS, together with sation of clumps resulting from gravita-
Although current results seem to re- the relatively large aperture of the ESO tional instabilities in the disk. How can
fute any strong generalized stellar pol- 3.6-m telescope, will play an important we distinguish these two scenarios?
lution among planet hosts, it is impor- role, since it will give the opportunity to One of the keys may come from the
tant to cross-check. One interesting obtain masses (or at least meaningful study of planets in binaries. The pres-
way of addressing this problem might upper limits) for the least massive planets ence of a stellar companion possibly
then be the study and comparison of detected by the photometric missions. plays an important role in the formation
the chemical contents of stars in visual of planets. It has been shown, for ex-
binary systems composed of similar Multiple Systems: Dynamical ample, that a stellar companion can
spectral type solar-type stars, one (or Interactions truncate the proto-planetary disk at a
both) of which having planetary-mass radius that depends mainly on the dis-
companions. Strong differences found Among the many planets that are ex- tance between the two stars. If so, and
would be interpreted as a sign that stel- pected to be found, some will surely be- considering the core accretion sce-
lar pollution is quite common. However, long to multi-planetary systems. Today, nario, we should not be able to see
only one clear case has been studied to only about 10 such cases are known, planets around stars members of bina-
date: the double visual system 16 Cyg but many stars that are already known ry systems that are closer than a given
A and B, where the B component is to harbour a planet also show system- limit. How close can a star have a com-
known to harbour a planet. This case atic trends in radial velocity, indicating panion and still have planets? The an-
does not show any clear difference in that at least a second companion is swer to this question is very important
the iron abundance, while a curious present in the system. While for the ma- to understand how giant planets are
lithium abundance difference is found. jority of cases this tendency might be formed.
simply due to the presence of low-mass
Transit Candidates stellar companions, in some others Concluding Remarks
they might be the telltale signatures of
Another important result of instru- a multi-planetary system. The gain in As the planet search programmes
ments like HARPS will be their ability to precision with instruments like HARPS are on their way, many more planetary
follow up planetary-like transit signa- will definitely permit us to search the al- companions are expected to be discov-
tures detected by photometry. ready known planet hosts for other ered in the next few years. Many hopes
There are currently more than 20 planet-mass companions, and to in- are now placed on instruments like
groups around the world trying to look crease the number of known multiple HARPS, that will provide radial veloci-
for signatures of the presence of plan- systems. ties of stars with a precision of 1 ms–1
ets around field dwarfs by looking for There is in fact much interesting in- or better. This will give us the opportu-
the brightness dimming as a putative formation that can come from these nity to dramatically improve the sam-
planet crosses its disk. In spite of the cases. Current results have shown that ples.
efforts, only very few results have been planets in multiple systems come fre- Other major contributions will come
announced, and none of these was quently in resonant orbits (see e.g. from future space missions like
confirmed to have a planetary origin. HD82943 – Figure 6). This is telling us COROT, Eddington, or Kepler, which
The only clear case of a real planetary a lot about the formation and migration will unveil thousands of short-period
transit detected so far was found for the of the exoplanets. planets around stars in the solar neigh-
37
bourhood. And, of course, the use of References Mayor M., Queloz D., 1995, Nature 378,
high-precision astrometric measure- 355.
ments with instruments like the VLTI or Boss A.P., 1997, Science 276, 1836. Pollack J.B., Hubickyj O., Bodenheimer P.,
the Keck interferometer will survey Charbonneau D., Brown T.M., Latham D.W., Lissauer J.J., Podolak M., Greenzweig Y.,
Mayor M., 2000, ApJ 529, L45. 1996, Icarus 124, 62.
“nearby” stars for long-period systems. Saar S.H., Donahue R.A., 1997, ApJ 485,
Goldreich P., Tremaine S., 1980, ApJ 241,
Altogether, these coming observational 425. 319.
facilities will definitely help us to con- Israelian G., Santos N.C., Mayor M., Rebolo Santos N.C., Israelian G., Mayor M., 2001,
struct a new and more complete view of R., 2001, Nature 411, 163. A&A 373, 1019.
how planetary systems are born and Jorissen A., Mayor M., Udry S., 2001, A&A Wolszczan A., Frail D. A., 1992, Nature 355,
how they evolve. 379, 992. 145.
1. Introduction of the Infrared Spectrometer and Array the deepest Ks-band in any field. The
Camera (ISAAC, Moorwood 1997), we first results are spectacular, demon-
Between October 1999 and October obtained ultradeep images in the Js strating the necessity of this deep NIR
2000 an undistinguished high-galactic (1.24 µm), H (1.65 µm) and Ks (2.16 imaging, and having direct conse-
latitude patch of sky, the Hubble Deep µm) bands. The combined power of an quences for our understanding of
Field South (HDF-S), was observed 8-metre-class telescope and the high- galaxy formation.
with the VLT for more than 100 hours quality wide-field imaging capabilities of The rest-frame optical light emitted
under the best seeing conditions. Using ISAAC resulted in the deepest ground- by galaxies beyond z ∼ 1 shifts into the
the near-infrared (NIR) imaging mode based NIR observations to date, and near-infrared. Thus, if we want to com-
pare 1 < z < 4 galaxies to their present-
day counterparts at similar intrinsic
wavelengths – in order to understand
their ancestral relation – it is essential
to use NIR data to access the rest-
frame optical. Here, long-lived stars
may dominate the total light of the
galaxy and the complicating effects of
active star formation and dust obscura-
tion are less important than in the rest-
frame ultraviolet. This therefore pro-
vides a better indicator of the amount of
stellar mass that has formed. Com-
pared to the selection of high-redshift
galaxies by their rest-frame UV light,
such as in surveys of Lyman Break
Galaxies (LBGs, Steidel et al. 1996a,b),
38
selection in the NIR Ks-band gives a
more complete census of the galaxies
that contribute to the stellar mass con-
tent of the early universe. And, by
studying these systems over a sub-
stantial redshift range, we can directly
see how they were assembled as we
look further back in time. Very deep
optical-to-infrared data in many filters
are required not only to access the
V606 – H
rest-frame optical light of galaxies and
constrain their stellar composition, but
also to determine the redshifts of
faint distant galaxies from their broad-
band photometry alone. The majori-
ty of the galaxies detected are too
faint to be observed spectroscopically,
even with powerful telescopes like
the VLT.
Here, we discuss the full NIR data set
of the HDF-S. The observations, reduc-
tion techniques, the catalogue of
sources, and the photometric redshifts
are described in detail in Labbé et al.
(2002). Throughout, we will assume a
flat Λ-dominated cosmology (ΩM = 0.3,
Λ = 0.7, H0 = 100h km s–1 Mpc–1) and
all magnitudes are expressed in the Figure 2: V606 – H versus H colour-magnitude diagram (on the AB system) for galaxies in the
Johnson photometric system unless HDFS Ks-selected catalogue with 1.95 < zphot < 3.5. Filled symbols indicate galaxies with
explicitly indicated by the subscript AB spectroscopy and the data points are grouped into two redshift ranges: 1.95 ≤ z < 2.7 (red)
for the AB magnitude system. and 2.7 ≤ z ≤ 3.5 (blue). We find 7 galaxies redder than V606 , AB – HAB > 3 and brighter than
HAB < 25, compared to only one in the HDF-N. Galaxies with S/N < 2 for the V606 measure-
2. Observations ment (triangles) are plotted at the 2σ confidence limit in V606, indicating a lower limit on the
V606 – H colour. Galaxies having red (Js – Ks) J > 2.3 colours (open squares) are also shown.
The number of candidates for red, evolved galaxies is much higher than in the HDF-N for a
The observations of the HDF-S are similar survey area, as shown in an identical plot in Figure 1 of Papovich, Dickinson &
part of the Faint InfraRed Extragalactic Ferguson (2001). The transformation of the V606 – Hs colour from the AB system to the
Survey (FIRES, Franx et al. 2000), a Johnson magnitude system is (V606 – H) J = (V606 – H)AB + 1.26.
large public programme carried out at
the VLT consisting of very deep NIR
ISAAC observations of two selected The colour image shown in Figure 1 and evolved galaxies with comparative-
fields with existing deep optical WFPC2 combines HST I-band data with ISAAC ly low star-formation rates, and models
imaging. The second and somewhat data in the Js and Ks bands. Most of galaxy formation must allow for their
shallower field around the z ≈ 0.83 clus- sources visible are very faint distant formation in sufficient numbers at z ∼ 3.
ter MS1054-03 (Förster Schreiber et al. galaxies and a rich variety in their opti- It appears that in recent hydrody-
in preparation) was observed for 80 cal-to-infrared colours is readily appar- namical simulations (Weinberg, Hern-
hours over an area four times larger ent. Some high-redshift galaxies, visi- quist & Katz 2002) high-redshift galax-
than the HDF-S. The full NIR data set ble as red sources, are hardly detected ies are converting gas into stars
of the HDF-S consists of 33.6 hours of or even absent in the optical I-band but continuously and at a fairly high rate,
imaging in Js, 32.3 hours in H, and 35.6 are quite bright in the NIR Ks. keeping them blue or “unevolved”.
hours in Ks with ISAAC, which has a While this is qualitatively consistent
2.5′ × 2.5′ field of view. The effective 3. Evolved Galaxies at High with the lack of red galaxies in the HDF-
seeing in the reduced images is ap- Redshift N, the situation in the HDF-S is signifi-
proximately 0.46″ in all bands, close to cantly different.
the best seeing that can be reasonably Remarkably, the HDF-S contains All high-redshift galaxies with very
obtained from Paranal. We reach a depth many high-redshift sources that are rel- red V606 – H colours also have very red
of 25.9 in Js, 24.8 in H, and 24.4 in Ks atively bright in the infrared and ex- infrared Js – Ks colours. Redshifting
(total magnitude for point sources, 3 σ). tremely faint in the optical, whereas the model spectra of galaxies shows that
Complemented by ultradeep HST/ HDF-N contains far fewer such galax- such red colours are produced by
WFPC2 imaging (version 2, Casertano ies. This is clearly visible in Figure 2, evolved systems at redshifts z > 2. In
et al. 2000) in the optical filters U300, which shows V606 – H colours versus our Ks-band selected catalogue we find
B450, V606, and I814 bands (the subscript the infrared H band magnitudes of NIR 13 relatively bright galaxies (Ktot < 22)
indicating the effective wavelength), we selected galaxies between redshifts 2 < that have red Js – Ks > 2.3 colours
assembled a Ks-selected catalogue z < 3.5. The filters are chosen to match (Franx et al. in preparation), compared
containing 833 sources, of which 624 those used in Figure 1 in Papovich, to 37 U-dropouts to the same flux limit.
have seven-band photometry, covering Dickinson & Ferguson (2001), allowing The photometric redshifts of these
0.3–2.2 µm. We determined the photo- a direct comparison with the HDF-N. galaxies, determined from the optical-
metric redshifts of all extragalactic For a similar survey area and similar to-infrared spectral energy distributions
sources using a method detailed in limiting depth, we find 7 galaxies redder (SEDs), are between 2 < zphot < 4. Yet,
Rudnick et al. (2001, 2002a in prepara- than V606,AB – HAB > 3 and brighter than most of the red galaxies are so faint in
tion). Comparison with spectroscopic HAB < 25 (on the AB magnitude sys- the observers optical that they would be
redshifts available for 49 sources in the tem), compared to one in the NIR sam- missed by standard optical selection
field shows excellent agreement: | zspec ple of HDF-N. These systems are prob- criteria, such as the U-dropout method.
– zphot | / (1+zspec ) ≈ 0.08. able candidates for relatively massive In principle, the red colours can be
39
Figure 3: Morphology of bright red Js – Ks > 2.3 galaxies in the HDF-S. The left grayscale panels show the averaged WFPC2 V+ I data, and
right-hand panels show our VLT/ISAAC Js, H, Ks data. The intensity is proportional to Fλ, with arbitrary normalization for each galaxy. The
colour images show a combination of I, Js, Ks data, after matching the image quality to that of the Ks-band. The right-hand column shows the
spectral energy distributions. Many of these galaxies are small and show prominent breaks in the infrared (rest-frame optical). Note that the
galaxy in the top row is barely visible even in Js. The SED of the object in the middle is consistent with a strong Balmer/4000Å break at z ∼
2.5. The galaxy in the bottom row is very extended in the rest-frame optical. It shows faint emission in the H-band out to 1″, consistent with
an exponential profile.
caused by dust, by contribution of 4. Strong Clustering of Faint haviour of Ks-selected galaxies at 2 <
prominent emission lines falling in the Red Galaxies at 2 < zphot < 4 zphot < 4, finding that the amount of
Ks-band, or by the redshifted Balmer or clustering depends strongly on the Js –
4000 Å break, which indicates the pres- Presumably, high-redshift galaxies Ks colour of galaxies , with red galaxies
ence of evolved stars. Although emis- with red Js – Ks colours are among the more clustered than blue galaxies, very
sion lines have been detected in spec- oldest and most massive galaxies at similar to what is observed in the opti-
troscopically confirmed Js – Ks > 2.3 their cosmic epoch and they have cal in the local universe. Dividing the
galaxies in the FIRES MS1054-03 field formed at the highest density peaks in sample at Js – Ks = 1.7 in two subsam-
by van Dokkum et al. (in preparation), the matter distribution at significantly ples reveals that the galaxies with Js –
their contribution to the broadband NIR earlier times. If so, this population Ks > 1.7 colours have the largest-ever
photometry is estimated to be small. In should be more clustered than less level of clustering measured for z > 2
many cases the spectral energy distri- massive and less evolved (bluer) ob- galaxies (see Fig. 4). The derived cor-
butions show a clear break between Js jects at similar redshifts. Using the relation length for the faint red galaxies
and Ks, which is more easily explained FIRES data of the HDF-S, Daddi et al. is r0 = 8.3 ± 1.2 Mpc (comoving): a fac-
as an aging effect than as a result of (2002) have studied the clustering be- tor of 3–4 higher than that of Lyman
dust reddening. These galaxies con-
tribute significantly to the rest-frame op-
tical luminosity density (Rudnick et al.
2002a in preparation) and from their Figure 4: The angu-
lar two-point correla-
red rest-frame optical colours – imply- tion function for
ing high mass-to-light ratios – we esti- galaxies with red Js
mate that they contain a substantial – Ks > 1.7 colours
fraction of the total stellar mass present between redshifts of
in all galaxies at z ∼ 3. 2 and 4, with the
The morphologies of the red galaxies best fitting power-law
are generally very compact at all wave- (solid line) and 1-
lengths, as can be seen in Figure 3. A sigma errors (dotted
notable exception is the galaxy in lines). A positive and
large clustering sig-
Figure 3c which is clearly extended and nal is detected, sug-
has an exponentially decreasing sur- gesting that these
face brightness profile out to 1″ radius red FIRES galaxies
in the H-band. This galaxy seems to are strongly
host blue knots which could form a spi- clustered in real
ral arm in the optical WFPC2 images, space, with a corre-
while it has an extended disk-like ap- lation length r0 = 8.3
pearance with a prominent bulge in the ± 1.2 h–1 Mpc
centre in the ISAAC H and Ks images; comoving. This is
the largest level of
possibly these are actively star-forming clustering ever found
sites embedded in a larger mature host for distant, z > 2
galaxy. galaxy populations.
40
Figure 5: Same as Figure 3 for the three largest Lyman-break galaxies in the HDF-S. The K-band images (rest-frame optical) are more cen-
trally concentrated than the WFPC2 images; this is a real change in light distribution and not caused by the decreased image quality in the
NIR. The redshifts of galaxies (a) and (c) have been confirmed spectroscopically; they have z = 2.027 and z = 2.793 respectively, implying a
scale length of 10h –1kpc for the latter.
Break galaxies over similar redshift these systems are very luminous giant deep optical observations to near-IR
ranges and with similar number densi- spiral galaxies, with patchy star forma- wavelengths for a more complete cen-
ties. The overall properties of this tion residing in a diffuse disk with a red sus of galaxies in the early universe. In
red J – K > 1.7 strongly clustered pop- bulge in the centre. It is well possible particular, the deepest-ever Ks-band
ulation suggests that they are the pro- that we have discovered the predeces- data have proven to be invaluable,
genitors of present-day massive ellipti- sors of the large disk galaxies we see probing well into the rest-frame optical
cals and extremely red objects (EROs) in the lower-redshift universe. Mor- at 2 < z < 4, where long-lived stars may
at z ∼ 1.5. phological studies in the rest-frame UV dominate the light of galaxies.
light of LBGs have always emphasized Provided with a new window on the
5. Large Disk-Like Galaxies at the compact and small sizes of the gal- early universe we find previously undis-
High Redshift axies (Giavalisco et al. 1996, Lowen- covered populations of galaxies, with
thal et al. 1997), as indeed is true for possible far-reaching consequences for
The high-redshift galaxies in HDF-S the bulk of our sample. However, galax- our understanding of galaxy formation.
show a large variation in both mor- ies that are large in the rest-frame optical We find that the HDF-S has many more
phologies and spectral energy distribu- with classical spiral morphologies have galaxies at z ∼ 3 with very red V606 – H
tions. Spectacularly, we find some z > 2 never been seen before, in particular not colours than the HDF-N; these are can-
galaxies that are very large in the rest- in the HDF-N (Dickinson et al. 2000) didates for relatively massive evolved
frame optical and show profound differ- even though such structures would systems. Closely connected, we find a
ences between the intrinsic ultraviolet have been well recognizable in their substantial population of red Js – Ks >
and optical morphologies (Labbé et al. deep WPFC2 and NICMOS imaging. 2.3 galaxies at z > 2, many of which are
in preparation). Three of the galaxies, Follow-up VLT/FORS spectroscopy barely detectable even in the deepest
shown in Figure 5, are LBGs (selected (Rudnick et al. 2002b in preparation) optical images, and would be missed
by applying the U-dropout criteria of confirmed the redshift of the galaxy in by optical colour selection techniques
Madau et al. 1996), while one is the Figure 5c, a disk-like U-dropout at zspec such as the U-dropout method. Yet,
previously described red galaxy in = 2.793 ± 0.003, implying a scale length these galaxies probably contribute sub-
Figure 3c. Their spectral energy distri- of 10 h –1kpc. Vanzella et al. (2002) found stantially to the total stellar mass pres-
butions show a pronounced break in an identical redshift for this galaxy and ent in the early universe at z ∼ 3. We
the rest-frame optical, identified as the for another, shown in Figure 5a, they detect strong clustering (r0 = 8.3 ± 1.2
age-sensitive Balmer/4000 Å break. found zspec = 2.027. Theories of disk Mpc, comoving) of galaxies with Js – Ks
They are amongst the brightest and formation generally predict that large >1.7 colours at 2 < zzphot < 4. The strong
reddest in the rest-frame optical and disk galaxies have been assembled re- clustering and their red Js – Ks colours,
they are probably the most massive cently (z < 1), and that high-redshift z > suggest that there is a direct evolution-
galaxies at z ∼ 2–3. 2 disks are small and dense, with typi- ary link between these systems, EROs
The I-band morphologies (probing cal sizes of ∼ few kpc. Producing large at z ∼1.5, and massive elliptical galax-
rest-frame 2000–2700 Å) are complex disk-like galaxies with reasonable likeli- ies in the local universe. We also find
with a patchy distribution that is not hood may therefore pose a challenge to high-redshift systems that are very
concentrated towards the centre. the current hierarchical models. large in the rest-frame optical with
Conversely, the NIR light is compact morphologies similar to those of
and peaks right in the centre of the I- 6. Summary nearby giant spiral galaxies, containing
band structure, surrounded by a diffuse a red bulge surrounded by a diffuse
and extended disk-like distribution. The These first results on the HDF-S disk and with scattered patches of star-
most straightforward explanation is that demonstrate the necessity of extending formation. While these are only a few
41
examples, it is currently not clear Acknowledgements Oke, J.B. 1971, ApJ, 170, 193.
whether galaxy formation models can Franx, M. et al., “FIRES at the VLT: the Faint
produce objects of comparable colours We wish to thank the ESO staff for InfraRed Extragalactic Survey”, The
and sizes in sufficient numbers to be operation of the VLT and ISAAC and Messenger 99, pp. 20–22, 2000.
consistent with our observations. their kind assistance. We are grateful Casertano, S. et al., AJ 120, pp. 2747–2824,
for their enormous efforts in obtaining 2000.
Lastly, the substantial differences Papovich, C., Dickinson, M., & Ferguson,
between HDF-N and HDF-S demon- these data in service mode and making
them available to us. H.C., ApJ 559, pp. 620–653, 2001.
strate that results based on such small Weinberg, D.H., Hernquist, L., & Katz, N.,
fields can be seriously affected by 2002, ApJ, 571, 15.
cosmic variance. The results of the References Daddi, E. et al., ApJ, submitted, 2002.
second field in the FIRES survey, the Moorwood, A.F.M., “ISAAC: a 1-5 µm imag- Madau, P., Ferguson, H.C., Dickinson, M.E.,
much larger MS1054-03 field, should er/spectrometer for the VLT”, in Optical Giavalisco, M., Steidel, C.C., & Fruchter,
decide which one of the fields is atypi- telescopes of today and tomorrow. A.L. A., MNRAS 283, pp. 1388–1404, 1996.
cal. We are pursuing extended follow- Ardeberg ed., proc. SPIE 2871, pp. Giavalisco, M., Steidel, C. C., & Macchetto,
up programmes to obtain more spec- 1146–1151, 1997. F.D., ApJ 470, 189, 1996.
troscopic confirmation of the above re- Steidel, C.C., Giavalisco, M., Dickinson, M., & Lowenthal, J.D. et al., ApJ 481, pp.
Adelberger, K.L., AJ 112, pp. 352-358, 1996. 673–688, 1997.
sults, allowing us to fully investigate
Steidel, C.C., Giavalisco, M., Pettini, M., Rudnick, G. et al., in preparation, 2002a.
the nature of these galaxies and the Rudnick, G. et al., in preparation, 2002b.
Dickinson, M., & Adelberger, K.L., ApJ
clues they provide for models of galaxy Dickinson, M., “The first galaxies: structure
462, L17, 1996.
formation. Updates on the FIRES pro- Rudnick, G. et al., AJ 122, pp. 2205–2221, and stellar populations” Philos. Trans. R.
gramme and access to the reduced im- 2001. Soc. London A 358, p. 2001, 2000.
ages and catalogues can be found at Labbé, I.F.L. et al., AJ accepted, 2002. Dickinson, M. et al., ApJ 531, pp. 624–634,
our website http://www.strw.leidenuniv.nl/ Förster Schreiber, N.M. et al., in preparation, 2000.
~fires. 2002. Vanzella, E. et al., A&A accepted, 2002.
By M. KISSLER-PATIG
Extragalactic Globular Clusters, tools for the study of galaxy formation by being superior in several practical
a Booming Field of Research and evolution. For this purpose they are aspects. For instance, globular clusters
used in nearby galaxies with as much are better chronometers than the dif-
Globular cluster systems were estab- success as the diffuse stellar popula- fuse stellar population since each glob-
lished in the last decade as powerful tions and complement the latter studies ular cluster can be identified as a sin-
gle-age population. The age determina-
tion of the major globular cluster sub-
populations allows one to precisely
date the star formation events in the
host galaxy.
Further, the study of globular cluster
systems is the easiest way to detect
multiple old or intermediate-age sub-
populations within a galaxy. These in
turn can trace multiple major star-for-
mation episodes at early times, invisi-
ble in the studies of the diffuse galaxy
light in which all populations are mixed.
Another advantage is that globular
clusters can be traced far out in the
halo, probing stellar populations and
kinematics at several effective radii of
the host galaxy – a range inaccessible
to diffuse stellar light studies. The glob-
ular clusters can be used for dynamical
studies at outer radii, shedding light on
Figure 1: NGC 6946 – a nearby spiral with a large number of very luminous young stellar clus-
the assembly histories of the systems
ters. Taken from Larsen’s contribution. The observations were made with the ALFOSC in-
strument on the Nordic Optical Telescope on La Palma. ALFOSC is a twin instrument of and the host galaxies.
DFOSC on the Danish 1.54-m and mounted on the NOT; it has a field size of about 6 × 6 arc- Finally, globular clusters are very
min. The colour image was generated from a mosaic of 3 pointings with red channel = V + I good calibrators at all wavelength for
+ Hα; green channel = V; blue channel = B. single stellar population models, an es-
42
sential tool for all galaxy formation and
evolution studies.
In summary, globular clusters are a
very strong complement to diffuse
galaxy-light studies. They are very
good tracers of the major star-formation
episodes of galaxies, including extreme
conditions such as star-burst during vi-
olent merger events. The study of these
systems allows us to learn not only
about the formation of the globular clus-
ters and their systems, but also about
the formation and evolution of galaxies.
43
1399 by Tom Richtler and his group), as Interestingly (even if not unexpected- old metal-poor globular clusters formed
well as on the chemistry of the old clus- ly), globular clusters are found in large prior to re-ionization.
ters (e.g. Puzia et al.). A number of numbers in the intergalactic medium Clearly, extragalactic globular clus-
studies also reported on large interme- of galaxy clusters. Michael Hilker and ters currently dominate the study of
diate age populations of globular clus- collaborators show clear evidence for stellar populations in nearby galaxies.
ters in early-type galaxies, detected globular clusters floating through the Their properties set hard constraints on
both by a combination of optical and Hydra I and Centaurus galaxy clus- galaxy formation and evolution models,
near-infrared photometry as well as by ters. but also shed new light on the star for-
spectroscopy (e.g. Hempel et al., Finally, globular clusters, as the old- mation history of the universe. This re-
Goudfrooij et al.). Interestingly, the est objects known in the sky, are now search area will clearly produce a number
globular clusters in dwarf galaxies look being put in a cosmological context. of astonishing results in the next 5 years.
“almost but not quite unlike” their count- Mike Beasley investigated the con- The organizers would like to warmly
er-parts in giant galaxies as Jennifer straints that they put on semi-analytic thank Christina Stoffer who perfectly
Lotz showed us. Despite the similarity, models in the hierarchical clustering controlled the logistics of the confer-
populations in dwarfs can extend to scenario of galaxy formation. Michael ence, as well as Pam Bristow for help-
very low metallicity. Santos presented a model in which the ing in the proceedings editing.
44
Celebrating ESO’s 40th Anniversary
C. MADSEN, ESO
ESO has always been a forward-
looking organization, focussed on pres-
ent and future programmes and proj-
ects. In its 40th year, this is as true as
ever, as ESO keeps its eyes on imple-
menting the VLTI, initiating the devel-
opment of the second-generation in-
strumentation for the VLT and, of
course, embarking on Phase II of the
ALMA project, for which the ESO
Council gave the green light at its mid-
year meeting in London. Still, a 40th an-
niversary warrants a look at the
decades passed and a reflection on the
role and achievements of the organiza-
tion during a period of time in which our
science has progressed tremendously.
That ESO has achieved its goals is
undisputed, providing evidence of the
wisdom of those astronomers who
gathered in Leiden in early 1954 to for-
mulate the need for this organization
and justifying the trust displayed by the Dr. Uwe Thomas addressing an invited audience during his visit to the ESO Headquarters on
five countries, which were the original 9 September.
signatories to the 1962 convention that
formed the legal base for ESO.
There were many milestones on September by Dr. Uwe Thomas, Secre- bound copies presented to Dr. Thomas,
ESO’s road to success. Not surprising- tary of State for education and research Dr. Freytag, Mr. Manfred Solbrig, the
ly, many of them were recalled – and of the German Federal Government. Mayor of Garching, and to ESO’s
retold – during the various celebrations Speeches by Dr. Thomas, Prof. Ralf Director General.
in September and October this year, Bender (LMU), Prof. Wolfgang Hille- On 14 October, a joint celebration by
described in the Anniversary issue of brandt (MPA), ESO’s Director General the Garching staff and the ESO Council
The Messenger, as well as, undoubted- and Dr. Arno Freytag, President of the took place, with speeches by ESO’s
ly, in numerous conversations among ESO Council, were followed by a Council President, the Director General
members of the ESO ‘family’, be it for- screening of a (pre-premiere) version of and Claudio Cumani, Chairman of the
mer or present staff, members of com- the anniversary film ‘Europe Reaches ESO International Staff Association.
mittees or Council – or users in the sci- for the Stars – Forty Years ESO’. Also, The event was ‘broadcast’ via ESO’s
entific community. the anniversary book ‘Geheimnisvolles video-conference system. The anniver-
Rather than organizing a single, Universum – Europas Astronomen sary film was shown to the entire staff,
high-profile central celebration event, entschleiern das Weltall’ by German both in Garching and at simultaneous
ESO opted for a series of events, be- science journalist Dirk Lorenzen was anniversary parties in Santiago, at La
ginning with the official visit on 9 shown for the first time with hand- Silla and Paranal.
As mentioned in the last issue of The
Messenger, ESO’s anniversary is also
a public event. Perhaps most visible is
the new planetarium show, produced
by the Association des planétariums de
langue française (APLF) in collabora-
tion with ESO. It is entitled ‘Mys-
teries of the Southern Sky’ and will be-
come available in French, German,
English, Dutch, Spanish and possibly
other languages. Until now, about 40
planetaria in France and seven other
countries have signed up and will stage
the show in the coming weeks and
months.
The formal premiere took place on 11
October at the Cité de l’Espace in
Toulouse. The premiere event for an in-
vited audience encompassed scientific
talks by Drs. Françoise Combes and
Marc Lachieze-Rey and an evening re-
ception accompanied by live Latin
American tunes.
Meanwhile, the ESO anniversary film
The newly refurbished auditorium in Garching was packed during the anniversary celebration is available on a three-language DVD
on 14 October. (English, French, German), as well as
45
In spite of the decidedly unpleasant weather in Garching on 14
October, good catering contributed to the convivial atmosphere in the
ESO’s Director General addressing the invited audience at the pre- party tent, as about 250 people gathered for the dinner for staff and
miere of the planetarium show in Toulouse. Council.
in single-language versions on VHS scheduled for early December. The some of the most spectacular astro-
tapes. Among other places, the DVD CD-ROM, which is produced in a col- nomical images based on observations
will be widely available through those laboration between ESO, USM (Mu- at La Silla and Paranal combined with
planetaria staging the ESO show. nich) and Planetary Visions (University advanced 3D simulation software. It will
Finally, the CD-ROM (in English) en- College, London), includes video clips, be commercially available all over
titled ‘3D Atlas of the Universe’, is now photos and texts about ESO as well as Europe.
Winds of Change
46
The EIROforum stand was packed with attendants as the EIROforum charter was signed.
inforced by the decision at the the dynamic created by the European model of the 100-m OWL telescope. It
Barcelona summit in March 2002 to aim Research Area. Europe has unques- caught the attention of many visitors
for a 50% increase in research spend- tioned excellence in science. By work- and was also shown in ‘The Sixth
ing in Europe, to reach the 3 % level of ing together, Europe’s leading research Sense’, the daily conference news-
the GDP by the year 2010. One of the organizations can make that more visi- paper. Apart from the OWL project,
most important elements in the effort to ble on the European and world stage,’ subjects covered on the ESO stand
revive European competitiveness in said Busquin in a press statement. were the VLT/VLTI and the ALMA
R&D is the concept of the European The signing of the charter was also project as well as those programmes
Research Area (ERA) and the 6th marked by the launch of the EIROforum for which the European Commission
Framework Programme is explicitly de- web site (www.eiroforum.org) and a has provided major financial support,
scribed as a tool to help facilitate the brochure describing the aims and the e.g. for Adaptive Optics, the Astro-
formation of the ERA. specific elements of the collaboration. physical Virtual Observatory (AVO) and
Clearly, creating the ERA is a com- the educational outreach programmes,
plex process with a diversity of actors EIROforum at the Conference notably ‘Life in the Universe’ and
and stakeholders. The conference pro- Indeed, EIROforum maintained a high ‘Physics on Stage’. Clearly, exposing
vided a forum for discussions among all visibility at the conference. Representa- ESO and its activities and future
groups involved, from science ministers tives of the member organizations took projects to such a huge, but select au-
to the scientists themselves, who part in several round-table discussions dience, was important, and the seven
turned up in huge numbers. and workshops, and the EIROforum ex- ESO staff that took turns on the stand
The subjects under discussion in- hibition attracted many visitors and pro- had many interesting exchanges with
cluded ‘Research and Innovation’, ‘Hu- vided inspiring surroundings for many the visitors.
man resources and mobility’, ‘Euro- discussions. The EIROforum stand cov- ‘European Research 2002’ sent a
pean research in a global context’, ered 400 square metres with an open- clear signal that the European re-
‘Infrastructures for the European Re- plan architecture, providing individual search landscape is undergoing a ma-
search Area’ as well as ‘Science and space for each of the seven member jor transformation, leading Le Figaro,
Society’. In parallel, ‘how-to’ seminars organizations and a centrally located the French daily, to talk about ‘a new
gave practical advice for scientists common area for the EIROforum itself. spirit for European scientists’. In two
planning to seek project funding under The press event in connection with the years time, when FP-6 reaches its
the new Framework Programme. signing of the EIROforum Charter took half-term, the follow-up meeting, an-
Furthermore, the ‘Participants’ forum’ place directly on the stand. nounced by Commissioner Busquin,
comprised more than 80 workshops will surely be able to determine to
and events covering a wide spectrum of The ESO Stand what extent this new spirit has pro-
topics – from fusion research, ge- Featuring prominently on the 36 duced tangible progress for European
nomics, ‘e-science’ and grid technolo- square metres ESO stand was a scale science.
gies to intellectual property rights and
science communication.
47
Agreement Between the Government of the
Republic of Chile and ESO for Establishing a New
Centre for Observation in Chile – ALMA
Reproduced from ESO Press Release 18/02 (23 October 2002)
48
There are many possibilities open to first Chilean telescope site. Contrac- instrumentation and observatory work,
a completing graduate student and it tually, we spend 80 nights a year (usu- which is a good complement to inde-
can be quite daunting filtering through ally in bite-size shifts of 1 week) on the pendent research. ESO fellowships are
the various options, academic and oth- mountain doing support work, plus up also very well regarded and presti-
erwise. For me, the decision was based to a month doing duty-related work in gious, so it is certainly a good position
on a few fundamental academic and Santiago. The rest of the time is your to have held.
personal priorities. I wanted a destina- own to pursue whatever scientific re- On the practical side, ESO fellows
tion that was going to provide an inter- search takes your fancy. The advan- are well paid, with allowances for being
esting and fulfilling lifestyle, preferably tage in this set-up is that your duty re- overseas and having to spend time on
with the chance to learn a new lan- sponsibilities can be quite easily com- the mountain, and the salary is essen-
guage. I was also looking for a working partmentalized: When you’re on the tially tax-free. There are also salary
environment that would offer diverse mountain you do support and when benefits if you are married and have
opportunities for professional develop- you’re in Santiago you do science. children (who additionally receive an
ment. Working in an observatory was For every week you spend on the education allowance). Taking into ac-
therefore very appealing, and at that mountain, you officially get the follow- count the lower cost of living in Chile,
time (as is still the case a few years lat- ing week off as ‘compensation time’. you can certainly live very well on the
er) there were plenty of positions as ob- However, most people tend to only take salary you take home, and there are
servatories expanded their horizons to a few days off in order to maximize the lots of opportunities to live well in Chile!
the new generation of large telescopes. time they can spend on their science, The lifestyle here is definitely an attrac-
In addition, as an undergraduate, I had as research output is always an impor- tion – Santiago has the beach, moun-
spent five consecutive summer vaca- tant factor in securing the next job. For tains, and vineyards on its doorstep
tions working at various observatories Chile fellows, the contracts are nomi- and all of the trappings of a big city
in locations including Australia, Hawaii, nally four years (a recent – and wel- such as cinemas, restaurants, and the-
and the Canary Islands, and undertak- come – change from the old style 3- atres.
ing support work and dabbling in instru- year contracts); the first three years are Of course, every job has its associat-
mentation seemed a good complement spent divided between research and ed drawbacks, and fellowships are no
to my scientific research. support whilst there are various options exception. The support work is very de-
Therefore, many of my postdoc ap- available for the final year. You might manding, requires a lot of commitment,
plications were aimed at observatories, continue working at ESO Chile, but with and involves a lot of travelling (it’s 6
and I was particularly excited by the only 40 nights of support work, or you hours door-to-door to Paranal) and
possibility of moving to Chile. The fel- might go back to the headquarters in night work. You also have to spend a
lowship application consists of a Germany and do a similar duty load in significant fraction of your time away
straightforward form (no long research one of the divisions there. Alternatively, from your family, and spouses do not
proposal like most places!) followed by you can elect to go to one of the have work status. Also, the support
an interview at the ESO headquarters Chilean universities where you will do work obviously means you can spend
near Munich in Germany with a video 100% research and have the benefits less time on science, although this is
link to Chile. Applicants are selected on of applying for telescope time as a something you have presumably ac-
the combined criteria of a strong scien- Chilean. Due to the relatively small cepted if you are considering an obser-
tific record and suitability for support number of Chilean astronomers com- vatory job. For me, I have actually
work (particularly important for Chile- pared to the large number of telescope found that my productivity has not seri-
bound fellows). Priority for posts goes hours they are entitled to, this offers a ously suffered and that time away at the
to nationals of ESO member countries, great opportunity to collect lots of data. telescope is a refreshing change from
and at that time the UK was not a mem- A final option that has just been ap- the office. The key to success is defi-
ber (it joined this July). However, be- proved is that fellows may return to nitely effective short-term time man-
cause Chilean fellows have to fulfill a university department that is willing agement; it is important to be able to
very specialist duties at the telescopes, to ‘adopt’ them in any ESO member plan your work on time scales of a few
there are fewer suitable applicants (and country with their salary still paid by weeks between each telescope shift.
usually fewer people who want to move ESO. This is a particularly attractive op- This means breaking down large tasks
to Chile) and this criterion is therefore tion for many as it gives fellows who into smaller jobs and setting goals and
relaxed. Because the ESO fellowship have been away from their home coun- deadlines on a weekly basis. Focusing
was my first choice amongst the vari- tries for three years the opportunity to in this way has allowed me to be much
ous positions I applied for, it was an advertise themselves to universities more efficient with my time when I’m
easy decision when, just before the that they see as potential employers not on the mountain, a technique that
start of the Christmas holidays, I was when their fellowship ends. This latter will be useful even when I’m no longer
offered a position to start the following option also offers the potential to gain working at ESO.
autumn. some teaching or supervisory experi- My contract with ESO will come to
Many people expect that moving to ence, opportunities that are limited at an end in September 2003. However, I
Chile may involve a large culture shock. ESO itself. already have a tenure-track job lined up
Although it is certainly not the same as The flexibility, diversity, and four at the University of Victoria in Canada,
Europe, the facilities and infrastructure guaranteed years are just a few of the allowing me to satisfy my ‘wanderlust’
in Santiago are very well developed advantages of the ESO fellowship still further and get to know another
and settling in was not a problem at all. scheme. Other professional pluses new country. Apart from my research
ESO has been dealing with Europeans include the opportunity to meet the profile, the experience and skills I have
moving to Chile for many years and has large number of European astronomers developed at ESO are definitely part of
a well-oiled system that handles every- that pass through to conduct their ob- what secured this position, as Canada
thing from moving your household and servations, providing ample possibili- increases its access to large tele-
getting your local ID and visa proc- ties for forming new collaborations and scopes and its involvement in future
essed to helping you find somewhere to making yourself known within the com- projects such as NGST and ALMA. I’ll
live and opening a bank account. Chile munity. In addition, it is generally at- certainly be sad to leave Chile though,
fellows are allocated duty stations at ei- tractive for future employers to see although I’m sure it will only be hasta
ther Paranal (like me) or La Silla, ESO’s that you have had experience with luego and not adios.
49
ANNOUNCEMENTS
FIRST ANNOUNCEMENT
ESO WORKSHOP ON
High Resolution Infrared Spectroscopy in Astronomy
ESO-Headquarters, Garching near Munich, November 18–21, 2003
Infrared spectroscopy at a resolution of a few km/s offers a unique tool to study rotational-vibrational transitions of many abundant mole-
cules as well as important atomic lines in a multitude of interesting astrophysical environments. Applications include the possible direct
detection of extrasolar planets, measurements of the abundances and magnetic fields of stars, studies of ISM chemistry and the kine-
matics of stars and gas in galactic centres.
The ESO VLT will shortly be equipped with two unique spectrometers which not only offer this high spectral resolution but also spatial res-
olutions of ~ 0.2″:
• CRIRES, an adaptive optics fed 1–5 µm spectrograph yielding a resolution of 3 km/s
• VISIR, which includes a high-resolution mode yielding a resolution of < 8 km/s between 8 and 13 µm
The aims of this workshop are to:
• present the latest status of the high-resolution infrared spectroscopic capabilities of the VLT and other observatories
• bring together the community interested in the application of high resolution infrared spectroscopy and to foster new
collaborations
• provide ESO with feedback in the phase when operating and data reduction software is being defined and coded
For space reasons the number of participants is limited to 110.
Scientific Organizing Committee:
Bengt Gustafsson (Uppsala, chair), Catherine de Bergh (Meudon), Ewine van Dishoeck (Leiden), Artie Hatzes (Tautenburg), Ken Hinkle
(Tucson), Ulli Käufl (ESO), Alan Moorwood (ESO, co-chair)
More details, preliminary pogramme and registration form can be retrieved from http://www.eso.org/gen-fac/meetings/ekstasy2003
or contact
[email protected] for further information
Symposium Secretary: Christina Stoffer
European Southern Observatory, Karl-Schwarzschild-Str. 2,
D-85748 Garching bei Muenchen, Germany
Fax: +49 89 3200 6480; E-mail [email protected]
50
CHILE LEIVA BECERRA, Alfredo, Instrumentation Engineer
DAHLEM, Michael (D), Operations Staff Astronomer ROA FIGUEROA, Luis, Mechanical Technician
DELSANTI, Audrey (F), Student ROJAS CHANAMPA, Chester, Instrument Technician
FAURE, Cécile (F), Student TORRES ZAMORANO, Manuel, Maintenance Electronic
ZALTRON, Paolo (I), Associate Technician
DEPARTURES
Local Staff (1 September – 30 November 2002) ALVAREZ CASTELLON, Pedro,
Precision Mechanic
ARRIVALS CASTRO CASTRO, Wilson, Mechanic
CANIGUANTE BARCO, Luis Alejandro, Electronic Engineer MIRANDA MANRIQUEZ, Marcela,
HENRIQUE LEON, Juan Pablo, Mechanical Technician Administrative Secretary
51
ESO, the European Southern Observa- No. Title Price
tory, was created in 1962 to “... establish
53 ESO/ST-ECF Workshop on “Calibrating and understanding HST and ESO
and operate an astronomical observato-
instruments”, Garching, Germany. P. Benvenuti (ed.) 7 30.–
ry in the southern hemisphere, equipped
with powerful instruments, with the aim of 54 Topical Meeting on “Adaptive Optics”, October 2–6, 1995, Garching,
furthering and organising collaboration Germany. M. Cullum (ed.) 7 40.–
in astronomy...” It is supported by ten 55 NICMOS and the VLT. A New Era of High Resolution Near Infrared Imaging
countries: Belgium, Denmark, France, and Spectroscopy. Pula, Sardinia, Italy, May 26–27, 1998 7 40.–
Germany, Italy, the Netherlands, Portu-
56 ESO/OSA Topical Meeting on “Astronomy with Adaptive Optics – Present
gal, Sweden, Switzerland and the United
Results and Future Programs”. Sonthofen, Germany, September 7–11,
Kingdom. ESO operates at two sites in
1999. D. Bonaccini (ed.) 7100.–
the Atacama desert region of Chile. The
new Very Large Telescope (VLT), the 57 Bäckaskog Workshop on “Extremely Large Telescopes”. Bäckaskog,
largest in the world, is located on Sweden, June 1–2, 1999. T. Andersen, A. Ardeberg, R. Gilmozzi (eds.) 7 30.–
Paranal, a 2,600 m high mountain ap-
proximately 130 km south of Antofa-
gasta, in the driest part of the Atacama
desert where the conditions are excellent
for astronomical observations. The VLT
consists of four 8.2-metre diameter tele-
Contents
scopes. These telescopes can be used
separately, or in combination as a giant TELESCOPES AND INSTRUMENTATION
interferometer (VLTI). At La Silla, 600 km
north of Santiago de Chile at 2,400 m L. Pasquini et al.: Installation and Commissioning of FLAMES, the VLT
altitude, ESO operates several optical Multifibre Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
telescopes with diameters up to 3.6 m F. Pepe, M. Mayor, G. Rupprecht: HARPS: ESO’s Coming Planet Searcher.
and a submillimetre radio telescope Chasing Exoplanets with the La Silla 3.6-m Telescope . . . . . . . . . . . . . . . 9
(SEST). Over 1300 proposals are made K. Kuijken et al.: OmegaCAM: the 16k ×16k CCD Camera for the VLT
each year for the use of the ESO tele-
scopes. The ESO headquarters are lo- Survey Telescope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
cated in Garching, near Munich, Ger- A. Glindemann: The VLTI – 20 Months after First Fringes . . . . . . . . . . . . . . . 18
many. This is the scientific, technical and B. Koehler, C. Flebus, P. Dierickx, M. Dimmler, M. Duchateau, P. Duhoux,
administrative centre of ESO where tech- G. Ehrenfeld, E. Gabriel, P. Gloesener, V. Heinz, R. Karban, M. Kraus,
nical development programmes are car- J.M. Moresmau, N. Ninane, O. Pirnay, E. Quertemont, J. Strasser,
ried out to provide the Paranal and La K. Wirenstrand: The Auxiliary Telescopes for the VLTI: a Status Report . . 21
Silla observatories with the most ad-
vanced instruments. There are also ex- Roberto Gilmozzi and Jason Spyromilio: Paranal Observatory – 2002 . . . . . . 28
tensive astronomical data facilities. ESO O. Hainaut: News from La Silla: Science Operations Department . . . . . . . . . . 30
employs about 320 international staff Two unusual views of La Silla . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
members, Fellows and Associates in
Europe and Chile, and about 160 local REPORTS FROM OBSERVERS
staff members in Chile.
N.C. Santos, M. Mayor, D. Queloz, S. Udry: Extra-Solar Planets . . . . . . . . . . 32
The ESO MESSENGER is published I. Labbé, M. Franx, E. Daddi, G. Rudnick, P.G. van Dokkum,
four times a year: normally in March,
June, September and December. ESO
A. Moorwood, N.M. Förster Schreiber, H.-W. Rix, P. van der Werf,
also publishes Conference Proceedings, H. Röttgering, L. van Starkenburg, A. van de Wel, I. Trujillo, and
Preprints, Technical Notes and other ma- K. Kuijken: FIRES: Ultradeep Near-Infrared Imaging with ISAAC of the
terial connected to its activities. Press Hubble Deep Field South . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Releases inform the media about partic-
ular events. For further information, con- OTHER ASTRONOMICAL NEWS
tact the ESO Education and Public
Relations Department at the following M. Kissler-Patig: Summary of the Workshop on Extragalactic Globular
address:
Cluster Systems hosted by the European Southern Observatory
EUROPEAN in Garching on August 27–30, 2002 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
SOUTHERN OBSERVATORY A. Glindemann: The VLTI: Challenges for the Future. Workshop
Karl-Schwarzschild-Str. 2 at Jenam 2002 in Porto . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
D-85748 Garching bei München P. Shaver and E. Van Dishoek: Summary of a Meeting on Science
Germany Operations with ALMA, held on Friday, 8 November 2002 . . . . . . . . . . . . . 44
Tel. (089) 320 06-0
C. Madsen: Celebrating ESO’s 40th Anniversary . . . . . . . . . . . . . . . . . . . . . . 45
Telefax (089) 3202362
[email protected] (internet) C. Madsen: Breaking the Ground for the European Research Area – The
URL: http://www.eso.org Conference ‘European Research 2002’ . . . . . . . . . . . . . . . . . . . . . . . . . . 46
http://www.eso.org/gen-fac/pubs/ A Nobel Prize for Riccardo Giacconi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
messenger/ Agreement Between the Government of the Republic of Chile and ESO for
Establishing a New Centre for Observation in Chile – ALMA . . . . . . . . . . . 48
Sara Ellison: Eighty Nights Up a Mountain . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
The ESO Messenger:
Editor: Peter Shaver
Technical editor: Kurt Kjär ANNOUNCEMENTS
Printed by First Announcement of an ESO Workshop on High-Resolution Spectroscopy
Universitätsdruckerei in Astronomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
WOLF & SOHN ESO Workshop on Large Programmes and Public Surveys . . . . . . . . . . . . . . 50
Heidemannstr. 166 Personnel Movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
D-80939 München Daniel Hofstadt: Manfred Ziebell Retires . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Germany
Jacques Breysacher: Christa Euler – Thirty-Seven Years of Service with ESO! 51
ISSN 0722-6691 New ESO Proceedings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
ESO Workshop Proceedings Still Available . . . . . . . . . . . . . . . . . . . . . . . . . . 51
52