Images online at http://www.eso.org/outreach/press-rel/pr-2003/pr-06-03.html

Summary

The initial commissioning period of the new HARPS spectrograph (High
Accuracy Radial Velocity Planet Searcher) of the 3.6-m telescope at the
ESO La Silla Observatory has been successfully accomplished in the period
February 11 – 27, 2003.

This new instrument is optimized to detect planets in orbit around other
stars (“exoplanets”) by means of accurate (radial) velocity measurements
with an unequalled precision of 1 meter per second. This high sensitivity
makes it possible to detect variations in the motion of a star at this
level, caused by the gravitational pull of one or more orbiting planets,
even relatively small ones.

“First Light” occurred on February 11, 2003, during the first night of
tests. The instrument worked flawlessly and was fine-tuned during
subsequent nights, achieving the predicted performance already during this
first test run.

The measurement of accurate stellar radial velocities is a very efficient
way to search for planets around other stars. More than one hundred
extrasolar planets have so far been detected, providing an increasingly
clear picture of a great diversity of exoplanetary systems.

However, current technical limitations have so far prevented the discovery
around solar-type stars of exoplanets that are much less massive than
Saturn, the second-largest planet in the solar system. HARPS will break
through this barrier and will carry this fundamental exploration towards
detection of exoplanets with masses like Uranus and Neptune.

Moreover, in the case of low-mass stars – like Proxima Centauri, cf. ESO
PR 05/03 – HARPS will have the unique capability to detect big “telluric”
planets with only a few times the mass of the Earth.

The HARPS instrument is being offered to the research community in the ESO
member countries, already from October 2003.

Captions: PR Photo 08a/03 and PR Photo 08b/03 show the HARPS spectrograph
during laboratory tests. The vacuum tank is open so that some of the
high-precision components inside can be seen. On PR Photo 08a/03, the
large optical grating by which the incoming stellar light is dispersed is
visible on the top of the bench; it measures 200 x 800 mm.

HARPS is a unique fiber-fed “echelle” spectrograph able to record at once
the visible range of a stellar spectrum (wavelengths from 380 – 690 nm) with
very high spectral resolving power (better than R = 100,000). Any light
losses inside the instrument caused by reflections of the starlight in the
various optical components (mirrors and gratings), have been minimised and
HARPS therefore works very efficiently.

First observations

Captions: PR Photo 08c/03 displays a HARPS untreated (“raw”) exposure of
the star HD100623, of the comparatively cool stellar spectral type K0V.
The frame shows the complete image as recorded with the 4000 x 4000 pixel
CCD detector in the focal plane of the spectrograph. The horizontal white
lines correspond to the stellar spectrum, divided into 70 adjacent
spectral bands which together cover the entire visible wavelength range
from 380 to 690 nm. Some of the stellar absorption lines are seen as dark
horizontal features; they are the spectral signatures of various chemical
elements in the star’s upper layers (“atmosphere”). Bright emission lines
from the heavy element thorium are visible between the bands – they are
exposed by a lamp in the spectrograph to calibrate the wavelengths. This
allows measuring any instrumental drift, thereby guaranteeing the
exceedingly high precision that qualifies HARPS. PR Photo 08d/03 displays
a small part of the spectrum of the star HD100623 following on-line data
extraction (in astronomical terminology: “reduction”) of the previous raw
frame, shown in PR Photo 08c/03. Several deep absorption lines are clearly
visible.

During the first commissioning period in February 2003, the high efficiency
of HARPS was clearly demonstrated by observations of a G6V-type star of
magnitude 8. This star is similar to, but slightly less heavy than our Sun
and about 5 times fainter than the faintest stars visible with the unaided
eye. During an exposure lasting only one minute, a signal-to-noise ratio
(S/N) of 45 per pixel was achieved – this allows to determine the star’s
radial velocity with an uncertainty of only ~1 m/s!. For comparison, the
velocity of a briskly walking person is about 2 m/s. A main performance goal
of the HARPS instrument has therefore been reached, already at this early
moment.

This result also demonstrates an impressive gain in efficiency of no less
than about 75 times as compared to that achievable with its predecessor
CORALIE. That instrument has been operating very successfully at the 1.2-m
Swiss Leonard Euler telescope at La Silla and has discovered several
exoplanets during the past years, see for instance ESO Press Releases (PR
18/98, PR 13/00 and PR 07/01). In practice, this means that this new planet
searcher at La Silla can now investigate many more stars in a given
observing time and consequently with much increased probability for success.

Extraordinary stability

Captions: PR Photo 08e/03 is a powerful demonstration of the extraordinary
stability of the HARPS spectrograph. It plots the instrumentally induced
velocity change, as measured during one night (9 consecutive hours) in the
commissioning period. The drift of the instrument is determined by
computing the exact position of the Thorium emission lines. As can be
seen, the drift is of the order of 1 m/s during 9 hours and is measured
with an accuracy of only 20 cm/s.

The goal of measuring velocities of stars with an accuracy comparable to
that of a pedestrian has required extraordinary efforts for the design and
construction of this instrument. Indeed, HARPS is the most stable
spectrograph ever built for astronomical applications. A crucial measure in
this respect is the location of the HARPS spectrograph in a climatized room
in the telescope building. The starlight captured by the 3.6-m telescope is
guided to the instrument through a very efficient optical fibre from the
telescope’s Cassegrain focus.

Moreover, the spectrograph is placed inside a vacuum tank to reduce to a
minimum any movement of the sensitive optical elements because of changes in
pressure and temperature. The temperature of the critical components of
HARPS itself is kept very stable, with less than 0.005 degree variation and
the spectrum therefore drifts by less than 2 m/s per night. This is a very
small value – 1 m/s corresponds to a displacement of the stellar spectrum on
the CCD detector by about 1/1000 the size of one CCD pixel, which is
equivalent to 15 nm or only about 150 silicon atoms! This drift is
continuously measured by means of a Thorium spectrum which is simultaneously
recorded on the detector with an accuracy of only 20 cm/s.

PR Photo 08e/03 illustrates two fundamental issues: HARPS performs with an
overall stability never before reached by any other astronomical
spectrograph, and it is possible to measure any nightly drift with an
accuracy never achieved before [1].

During this first commissioning period in February 2003, all instrument
functions were tested, as well as the complete data flow system hard- and
software. Already during the second test night, the data-reduction pipeline
was used to obtain the extracted and wavelength-calibrated spectra in a
completely automatic way. The first spectra obtained with HARPS will now
allow the construction of templates needed to compute the radial velocities
of different types of stars with the best efficiency.

The second commissioning period in June will then be used to achieve the
optimal performance of this new, very powerful instrument. Astronomers in
the ESO community will have the opportunity to observe with HARPS from
October 1, 2003.

Other research opportunities opening

This superb radial velocity machine will also play an important role for the
study of stellar interiors by asteroseismology. Oscillation modes were
recently discovered in the nearby solar-type star Alpha Centauri A from
precise radial velocity measurements carried out with CORALIE (see ESO PR
15/01). HARPS is able to carry out similar measurements on fainter stars,
thus reaching a much wider range of masses, spectral characteristics and
ages.

Michel Mayor, Director of the Geneva Observatory and co-discoverer of the
first known exoplanet, is confident: “With HARPS operating so well already
during the first test nights, there is every reason to believe that we shall
soon see some breakthroughs in this field also”.

The HARPS Consortium

HARPS has been designed and built by an international consortium of research
institutes, led by the Observatoire de Geneve (Switzerland) and including
Observatoire de Haute-Provence (France), Physikalisches Institut der
Universitaet Bern (Switzerland), the Service d’Aeronomie (CNRS, France), as
well as ESO La Silla and ESO Garching.

The HARPS consortium has been granted 100 observing nights per year during a
5-year period at the ESO 3.6-m telescope to perform what promises to be the
most ambitious systematic search for exoplanets so far implemented
worldwide.

The project team is directed by Michel Mayor (Principal Investigator),
Didier Queloz (Mission Scientist), Francesco Pepe (Project Managers
Consortium) and Gero Rupprecht (ESO representative).

More information

For more details about the HARPS Spectrograph, please consult the HARPS
websites at the Geneva Observatory and at ESO, or the articles about this
new instrument in the ESO Messenger Nos. 105 and 110.

Note

[1] The high level of stability achieved by the thermostatically controlled
vacuum-spectrograph HARPS has never before been reached by any other
astronomical spectrograph. In the case of more conventional instruments,
drifts of several hundreds of m/s may occur during one observing night due
to the variation of atmospheric pressure (at the rate of about 90 m/s for 1
mbar variation) or the ambient air temperature (300 m/s for 1n).
It is expected that there will be a smooth drift of perhaps a few tens of
m/s during one year, due to thermo-mechanical flexure and relaxation of the
HARPS instrument. However, since such a long-term drift can be measured
regularly and very accurately by means of the Thorium calibration spectrum,
this effect can be fully compensated for and thus poses no problem. In fact,
the measured nightly drift of HARPS is so small and so smooth that it is
possible to compute an average value of this drift with an accuracy of only
a few cm/s.

Contacts

Didier Queloz
Observatoire de Geneve
Sauverny, Switzerland
Phone: +41 22 755 26 11
email: didier.queloz@obs.unige.ch

Gero Rupprecht
European Southern Observatory
Garching, Germany
Phone: +49-89-3200-6355
email: grupprec@eso.org