You are on page 1of 72

ELT M4 — The Largest Adaptive Mirror Ever Built

A Celebration of GRAVITY Science

The ESO Summer Research Programme 2019
The Messenger
No. 178 – Quarter 4 | 2019
ESO, the European Southern Observa- Contents
tory, is the foremost intergovernmental
astronomy organisation in Europe. It is Telescopes and Instrumentation
supported by 16 Member States: Austria, Vernet E. et al. – ELT M4 — The Largest Adaptive Mirror Ever Built 3
­Belgium, the Czech Republic, Denmark, Kasper M. et al. – NEAR: First Results from the Search for Low-Mass ­
France, Finland, Germany, Ireland, Italy, Planets in a Cen 5
the Netherlands, Poland, Portugal, Spain, Arnaboldi M. et al. – Report on Status of ESO Public Surveys and
Sweden, Switzerland and the United Current Activities 10
Kingdom, along with the host country of Ivanov, V. D. et al. – MUSE Spectral Library 17
Chile and with Australia as a Strategic
Partner. ESO’s programme is focussed GRAVITY Science
on the design, construction and opera- GRAVITY Collaboration – Spatially Resolving the Quasar Broad Emission
tion of powerful ground-based observing Line Region 20
­facilities. ESO operates three observato- GRAVITY Collaboration – An Image of the Dust Sublimation Region in the
ries in Chile: at La Silla, at P
­ aranal, site of Nucleus of NGC 1068 24
the Very Large Telescope, and at Llano GRAVITY Collaboration – GRAVITY and the Galactic Centre 26
de Chajnantor. ESO is the European GRAVITY Collaboration – Spatially Resolved Accretion-Ejection in
­partner in the Atacama Large Millimeter/ Compact Binaries with GRAVITY 29
submillimeter Array (ALMA). Currently GRAVITY Collaboration – Images at the Highest Angular Resolution
ESO is engaged in the construction of the with GRAVITY: The Case of h Carinae 31
Extremely Large ­Telescope. Wittkowski M. et al. – Precision Monitoring of Cool Evolved Stars:
Constraining Effects of Convection and Pulsation 34
The Messenger is published, in hardcopy GRAVITY Collaboration – Multiple Star Systems in the Orion Nebula 36
and electronic form, four times a year. GRAVITY Collaboration – Probing the Discs of Herbig Ae/Be Stars at
ESO produces and distributes a wide Terrestrial Orbits 38
variety of media ­connected to its activi- GRAVITY Collaboration – Spatially Resolving the Inner Gaseous Disc of the
ties. For further information, including Herbig Star 51 Oph through its CO Ro-vibration Emission 40
postal subscription to The Messenger, Davies C. L. et al. – Spatially Resolving the Innermost Regions of the
contact the ESO Department of Commu- Accretion Discs of Young, Low-Mass Stars with GRAVITY 43
nication at: Dong S. et al. – When the Stars Align — the First Resolved Microlensed Images 45
GRAVITY Collaboration – Hunting Exoplanets with Single-Mode
ESO Headquarters Optical Interferometry 47
Karl-Schwarzschild-Straße 2
85748 Garching bei München, Germany Astronomical News
Phone +498932006-0 Christensen L. L., Horálek P. – Light Phenomena Over ESO’s Observatories IV: Dusk and Dawn 51
Manara C. F. et al. – The ESO Summer Research Programme 2019 57
The Messenger Boffin H. M. J. et al. – Report on the ESO Workshop
Editor: Gaitee A. J. Hussain “Artificial Intelligence in Astronomy” 61
Layout, Typesetting, Graphics: Vieser W. et al. – Report on the IAU Conference
Jutta B
­ oxheimer, Mafalda Martins “Astronomy Education — Bridging Research & Practice” 63
Design, P­ roduction: Jutta ­Boxheimer Kokotanekova R., Facchini S., Hartke J. – Fellows at ESO 67
Proofreading: Peter Grimley In Memoriam Cristian Herrera González 70
­w Personnel Movements 71
Patat F. – Erratum: The Distributed Peer Review Experiment 71
Printed by FIBO Druck- und Verlags GmbH
Fichtenstraße 8, 82061 Neuried, Germany

Unless otherwise indicated, all images in

The Messenger are courtesy of ESO,
except authored contributions which are
courtesy of the respective authors.
Front cover: Simulation of the orbits of stars very
close to the supermassive black hole at the heart of
© ESO 2019 the Milky Way, Sgr A*. One of these stars, S2, is the
ISSN 0722-6691 perfect laboratory to test Einstein’s general theory
of relativity as it passes very close to the black hole,
with an orbital period of 16 years. S2’s orbit has
been monitored with ESO’s telescopes since the
1990’s and continues at even greater precision with
GRAVITY. Credit: ESO/L. Calçada/

2 The Messenger 178 – Quarter 4 | 2019

Telescopes and Instrumentation DOI: 10.18727/0722-6691/5162

ELT M4 — The Largest Adaptive Mirror Ever Built

Elise Vernet 1 (approximately a ninth of the full moon).

Michele Cirasuolo 1 Thanks to the combined use of M4 and
Marc Cayrel 1 M5, the optical system is capable of
Roberto Tamai 1 ­correcting for atmospheric turbulence
Aglae Kellerer 1 and the vibration of the telescope struc-
Lorenzo Pettazzi 1 ture itself induced by motion and wind.
Paul Lilley 1
Pablo Zuluaga 1 This adaptive capability is crucial to
Carlos Diaz Cano 1 allowing the ELT to reach its diffraction
Bertrand Koehler 1 limit, which is ~ 8 milliarcseconds (mas) in
Fabio Biancat Marchet 1 the J-band (at λ ~ 1.2 μm) and ~ 14 mas
Juan Carlos Gonzalez 1 in the K-band. In so doing the ELT will
Mauro Tuti 1 be able to yield images 15 times sharper
+ the ELT Team than the Hubble Space Telescope and
with much greater sensitivity. Translated
into astrophysical terms this means
ESO opening up new discovery spaces, from Figure 1. Rendering of the M4 adaptive mirror unit
for the ELT.
exoplanets closer to their stars, to black
holes, to the building blocks of galaxies
The Extremely Large Telescope (ELT) is both in the local Universe and billions of consortium name of AdOptica. Many
at the core of ESO’s vision to deliver the light years away. For example, the ELT 8-metre telescopes now have a metre-
largest optical and infrared telescope will be able to detect and characterise scale adaptive mirror. The same tech­
in the world. Continuing our series of extrasolar planets in the habitable zone nology is now being adapted to serve the
Messenger articles describing the opti- around our closest star Proxima Centauri, ELT, in order to produce a mirror with an
cal elements of the ELT, we focus here or to resolve giant molecular clouds (the area five times larger. The M4 mirror uses
on the quaternary mirror (M4), a true building blocks of star formation) down to the same principle as a loudspeaker; the
technological wonder; it is the largest ~ 50 parsecs in distant galaxies at z ~ 2 mirror is made of a very thin shell levitating
deformable mirror ever made. In combi- (and even smaller structures for sources 100 microns away from its reference sur-
nation with M5, M4 is vital to delivering that are gravitationally lensed by fore- face (this corresponds to the thickness
the sharp (diffraction-limited) images ground clusters) with an unprecedented of a standard A4 sheet of paper) and it
needed for science by correcting for sensitivity. acts like a membrane which deforms
atmospheric turbulence and the vibra- under the effect of about 5000 voice coil
tions of the telescope itself. Here we actuators. A voice coil actuator is a type
describe the main characteristics of M4, The quaternary mirror (M4) of direct drive linear motor and the name
the challenges and complexity involved “voice coil” comes from one of its first
in the production of this unique adaptive M4 is the main adaptive mirror of the tele- historical applications, vibrating the paper
mirror, and its manufacturing status. scope. The term “adaptive mirror” means cone of a loudspeaker. It consists of a
that its surface can be deformed to cor- permanent magnetic field assembly and
rect for atmospheric turbulence, as well a coil assembly. The current flowing
Background: how the ELT works as for the fast vibration of the telescope through the coil assembly interacts with
structure induced by its motion and the the permanent magnetic field and gener-
Let’s briefly recall how the ELT works. wind. In the case of M4, more than 5000 ates a force that can be reversed by
The optical design of the ELT is based on actuators are used to change the shape changing the polarity of the current.
a novel five-mirror scheme capable of of the mirror up to 1000 times per second.
collecting and focusing the light from Depending on the current injected into
astronomical sources and feeding state- In combination with the M5 mirror, M4 the coil the mirror can be pushed or
of-the-art instruments for the purposes of forms the core of the adaptive optics of pulled up to a distance of 90 microns
imaging and spectroscopy. The light is the ELT. With a diameter of 2.4 metres, from its mean position. With the help of
collected by the giant primary mirror M4 will be the largest adaptive mirror ever a very fast and precise set of capacitive
39 metres in diameter, relayed via the M2 built. By comparison, current adaptive sensors and amplifiers that are co-located
and M3 mirrors (each of which has a mirrors are just over 1 metre in diameter, with the voice coil actuators, the mirror’s
diameter of ~ 4 metres) to the M4 and M5 for example the 1.1-m M2 adaptive sec- position is measured 70 000 times per
mirrors that form the core of the adaptive ondary on the VLT UT4 telescope (Yepun). second to an accuracy of a few tens of
optics of the telescope; the light then nanometres (the size of the smallest virus)
reaches the instruments on one or other Adaptive mirror technology was trans- with the actuators being driven up to
of the two Nasmyth platforms. This lated into an industrial product for astron- 1000 times per second.
design provides an unvignetted field of omy more than two decades ago by
view (FoV) of 10 arcminutes in diameter the Italian companies Microgate s.r.l and M4 is made of several state-of-the-art
on the sky, ~ 80 square arcminutes ADS, internationally known under the components, the mirror and its reference

The Messenger 178 – Quarter 4 | 2019 3

Telescopes and Instrumentation Vernet E. et al., ELT M4 — The Largest Adaptive Mirror Ever Built

Figure 2. (Left) One of

the shell mirrors of the
M4 in Z
­ erodur©.

Figure 3. (Right) The

­reference body in
silicon-­c arbide being
inspected after brazing
the six parts.

structure being two of the most critical The back of the reference structure is

Mersen Boostec
ones. The mirror is an assembly of six supported by a 12-point whiffletree and
optically polished thin shells, or petals, laterally at six points on the mirror edge.
made of the low-expansion glass-ceramic The overall M4 sub-system is mounted
Zerodur© (manufactured by Schott on six position actuators (a hexapod sys-
GmbH). The six petals are obtained from tem), which provide the fine alignment
a 35 mm-thick blank, which is polished of the mirror. It is further mounted on a
and thinned down to a thickness of less rotating mechanism (called a switcher)
than 2 mm — necessary to achieve the which is used to select the Nasmyth
desired flexibility for shaping the mirror — focus to which the light will be directed.
and then finally cut into a precise shape
by Safran Reosc (France; see Figure 2).
Manufacturing the M4
In order to adjust the shapes of the thin
shells, a rigid and sufficiently accurate Safran Reosc (France) started to manu-
flat reference structure is also needed to facture the thin segment mirrors in 2017
hold the petals. This structure must be and four thin shells are now ready for Figure 4. Detail of the M4 reference body.
stiff enough to provide a good reference integration in Italy. The remaining eight
surface, whatever the orientation of the shells still need to be delivered in order
telescope. It also needs to hold all the to have two sets of six shells each (during The final integration will start at AdOptica
actuators, which will deform and change ELT operation one set is integrated on once the reference structure has been
the shape of the six petals. M4, while the other is being recoated). delivered. Given the number of compo-
The reference body manufacturing also nents that need to be assembled to a
The 2.7-metre diameter lightweight began in 2017 and six segments have high degree of precision, the integration
­structure is made of Boostec® silicon been brazed in the last few months. The will be a lengthy task requiring proce-
carbide, one of the stiffest materials reference surface will need to be lapped dures to ensure that the assembly and
­available (stiffer than steel, carbon fibre to 5 microns flatness before being deliv- calibration meet requirements. It should
or beryllium). Its surface has more than ered to Italy. take 1.5 years to fully integrate the M4
5000 holes which will hold the actuators mirror and start the final calibration of
(see Figure 4), while the back surface is To have a mirror fully tested in Chile by each mirror segment and their associated
composed of several ribs to reinforce the early 2024, AdOptica has to ensure capacitive sensors. A test tower is being
structure. Owing to its large dimensions, the procurement and manufacture of all specially developed to verify and test the
the silicon carbide structure is made of the other components, including all the M4. It will be used in Europe to calibrate
six parts brazed together, similar to the voice coil actuators and more than 60% the M4 unit before being transferred
Herschel primary mirror which was man- of the permanent magnets, which are to Chile where it will be used before the
ufactured more than a decade ago. The already in house and are waiting to be mirror is installed on the t­elescope and
manufacture of the structure is signifi- integrated. In addition, more than half of kept on-site for any future major mainte-
cantly challenging, not only because of the electronics boards are either ready nance activities that may be required.
the depth, length, and thickness of the or under calibration, and most of the
ribs, but also given the requirements on mechanical parts are ready, including the
its straightness, as well as the number reference structure cell support and its
and accuracy of the actuator holes. whiffletree.

4 The Messenger 178 – Quarter 4 | 2019

Telescopes and Instrumentation DOI: 10.18727/0722-6691/5163

NEAR: First Results from the Search for Low-Mass

­Planets in a Cen

Markus Kasper 1 ESO, in collaboration with the Break- ESO 3.6-metre telescope at La Silla,
Robin Arsenault 1 through Initiatives, has modified the VLT was modified and used to carry out the
Ulli Käufl 1 mid-infrared imager VISIR to greatly acceptance tests of the internal chopper.
Gerd Jakob 1 enhance its ability as a planet finder. It This was followed by a performance eval-
Serban Leveratto 1 has conducted a 100-hour observing uation of the Annular Groove Phase Mask
Gerard Zins 1 campaign to search for low-mass plan- (AGPM) coronagraph with a dedicated
Eric Pantin 2 ets around both components of the optical setup incorporating a line-tunable
Philippe Duhoux 1 binary a Centauri, part of the closest CO2 laser, elliptical mirrors and germa-
Miguel Riquelme 1 stellar system to the Earth. Using adap- nium lenses. Four AGPM coronagraphs
Jean-Paul Kirchbauer 1 tive optics and high-performance coro- were tested, three specifically optimised
Johann Kolb 1 nagraphy, the instrument reached for the NEAR filter (10–12.5 μm) and an
Prashant Pathak 1 unprecedented contrast and sensitivity older sample manufactured in 2012 and
Ralf Siebenmorgen 1 allowing it to see Neptune-sized planets optimised for wavelengths between 11
Christian Soenke 1 in the habitable zone, if present. The and 13.1 μm. Surprisingly, the older coro-
Eloy Fuenteseca 1 experiment allowed us to characterise nagraph performed best, with a rejection
Michael Sterzik 1 the current limitations of the instrument. ratio of up to 400 at 10.5 μm, and a con-
Nancy Ageorges 3 We conclude that the detection of trast level of < 10 – 4 at 3 λ/D.
Sven Gutruf 3 rocky planets similar to Earth in the
Dirk Kampf 3 habitable zone of the a Centauri System After passing Provisional Acceptance
Arnd Reutlinger 3 is already possible with 8-metre-class Europe (PAE) in November 2018, the
Olivier Absil 4 tele­scopes in the thermal infrared. NEAR hardware was shipped to Paranal.
Christian Delacroix 4 At the same time, VISIR was dismounted
Anne-Lise Maire 4 from UT3 (Melipal) and brought to Para-
Elsa Huby 5 From an idea to the telescope nal’s New Integration Hall (NIH) in prepa-
Olivier Guyon 6, 7 ration for the on-site installation starting
Pete Klupar 7 The a Centauri system is uniquely suited in early January 2019. As expected, three
Dimitri Mawet 8 to the search for signatures of low- cool-downs of VISIR were required to
Garreth Ruane 8 mass planets in the thermal infrared. The successfully implement all the new modi-
Mikael Karlsson 9 N-band at around 10 μm is best suited fications. First, the aperture wheel was
Kjetil Dohlen 10 for such observations, because this rearranged with the help of the Paranal
Arthur Vigan 10 is where a planet with a temperature like mechanical workshop to include two new
Mamadou N’Diaye 11 Earth’s is brightest. The a Centauri AGPMs and a special optical mask
Sascha Quanz 12 binary consists of the solar-type stars (ZELDA, N’Diaye et al., 2014) to measure
Alexis Carlotti 13 a Centauri A and B, and the planet-­ and pre-compensate optical aberrations
hosting (Anglada-Escudé et al., 2016) in the instrument. New Lyot filters were
M-dwarf star Proxima Centauri. In a mounted and mechanically centred with
ESO ­previous Messenger article (Kasper et al., the cold stop of VISIR to an accuracy of
AIM, CEA, CNRS, Université Paris- 2017), we provided details of how we better than 175 μm (i.e., 1% of the pupil
Saclay, Université Paris Diderot, planned to modify the existing VISIR diameter). The internal chopper, the
Sorbonne Paris Cité, Gif-sur-Yvette, instrument to conduct the necessary wavefront sensor arm and the calibration
France observations with the Very Large Tele- unit were installed with the help of the
Kampf Telescope Optics (KT Optics), scope (VLT). This article describes how contractor KT Optics, and all units were
Munich, Germany VISIR was moved to UT4, the innova­- successfully tested. In particular, the
University of Liège, Liège, Belgium tions and new technologies that were alignment of the calibration unit, which
Observatoire de Paris-Meudon, France implemented and how they work, con- uses an elliptical mirror with an aberration-­
Subaru Telescope, Tokyo, Japan cluding with the execution of the NEAR free field of view of around 0.1 mm in
Breakthrough Initiatives, Mountainview, (New Earths in the a Centauri Region) diameter was laborious and required
USA experiment — a unique 100-hour obser- some modifications of the mechanical
Caltech, Pasadena, USA vation of the a Centauri system, which mounts on-site.
Uppsala University, Sweden took place in early June 2019.
L aboratoire d’Astrophysique Marseille, Following the completion of the assembly
France Three years were needed to develop the integration and verification (AIV) activities,
Observatoire de la Côte d’Azur, Nice, NEAR experiment from the initial idea, VISIR was transported and mounted to
France from the Phase A review held in July 2016 UT4 (Yepun) in mid-March 2019 (see Fig-
Eidgenössische Technische Hochschule to the observing campaign in June 2019. ure 1). After measuring the expected
Zürich, Switzerland Between January and July 2018, ESO’s residual misalignment between the instru-
Institude de Planétologie et d’Astro­ mid-infrared detector test facility Thermal ment and telescope pupil on-sky on
physique de Grenoble, France Infrared MultiMode Instrument (TIMMI2), 24 March, VISIR was taken off the tele-
a decommissioned instrument from the scope again for adjustment by tilting

The Messenger 178 – Quarter 4 | 2019 5

Telescopes and Instrumentation Kasper M. et al., NEAR: First Results from the Search for Low-Mass P
­ lanets in a Cen

the instrument, and some fine adjustment Figure 1. (Left) VISIR

ESO/NEAR Collaboration
mounted on UT4 and
of the wavefront sensor arm. On-sky
ready for NEAR. The
commissioning started on 3 April 2019 alternative altitude cable
and lasted for 10 half-nights, during wrap connecting the
which the various new functions were instrument to the elec-
tronics racks and helium
tested, and operational procedures were
compressors on the
tuned. ­a zimuth platform can be
seen on the left hanging
down from the mirror
Technical innovations, observing modes
and performance

NEAR implements several technologies

which are either completely new for
N-band astronomy or have not previously
been tested on-sky at this wavelength.
For example, the experiment confirmed
that atmospheric water vapour content
does not significantly impact the adaptive
optics (AO) corrected N-band image
quality, and that mid-infrared spectral fil-
ters can be overcoated with chromium
masks implementing Lyot stops or apo-
disers for the coronagraph. We also, for
the first time, implemented an alternative
altitude cable wrap (see Figure 1), which
could also greatly facilitate the operation
of other Cassegrain instruments.
Figure 2. (Below) Illus-
tration of the VISIR data
acquisition of a Centauri
Chopping, internal and external with chopping.

Among the new technologies is an internal

chopping device, the so-called Dicke
a Cen B a Cen B
Switch, which is described in more detail
in Kasper et al. (2017). We tested the
Dicke Switch at chopping frequencies
a Cen A – a Cen B
up to 10 Hz during commissioning, and AGPM and WFS (on AGM coro)
it substantially reduces the detector’s
Excess Low Frequency Noise (ELFN) as
a Cen A
foreseen. There is an expected mismatch a Cen A
in the spatial distribution of the sky and Chop A Chop B Chop A – Chop B
internal background, but this mismatch
turns out to be stable in time and can be
well modelled or subtracted by nodding (SPARTA)2 made sure that DSM chopping the left and middle panels, and the
techniques. This device can be used when observations are highly efficient and ­chopping subtracted image of the two
external chopping is not possible — when, almost transparent to the instrument. In on the right.
for example, the source size exceeds the addition, the a Centauri binary offers
throw range of an external chopper. the possibility of chopping with an ampli-
tude corresponding to the separation Coronagraph modes and centring
The second option, external chopping between the two stars of about 5 arcsec-
using the Deformable Secondary Mirror onds in 2019, placing all the time a scien- The light from the star at the location
(DSM), worked flawlessly. This option tifically interesting target on the corona- where we search for planets can be sup-
was initially deemed a risky approach, graphic mask and doubling the efficiency. pressed using two different concepts in
because the chopping action is seen by Because of these advantages, we used NEAR. The first is the AGPM, a technical
the AO and could have disturbed its external chopping with the DSM for the realisation of a Vortex coronagraph using
operation. However, the clever design of a Centauri observations, and Figure 2 a sub-wavelength grating etched into a
the DSM and the Standard Platform for illustrates the data as seen by the detec- diamond substrate (Mawet et al., 2005).
Adaptive optics Real Time Applications tor during the two chopping cycles on The second is a shaped pupil mask

6 The Messenger 178 – Quarter 4 | 2019

0.10 9 120
Estimated pointing error (λ/D)

y 8
0.05 7

Cumulative time (h)

6 80

Time/night (h)
0.00 5
– 0.05 3 40
– 0.10 20
0 0
0.25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Relative field selector position (arcseconds)

Nights of 23 May–11 June, 26 June

Time/night (h) Cumulative time (h)
Figure 3. (Left) Top: The shifting move- Figure 4. (Above) Hours of open shutter time over
ment of the star behind the AGPM is the duration of the NEAR campaign. Only small
0.15 measured by the QACITS algorithm in amounts of data could be collected between 24 and
a closed loop during a NEAR observ- 28 May and between 5 and 11 June, owing to medio-
ing night. The horizontal axis shows cre (mostly cloudy) observing conditions.
the hour angle and the colours refer to
0.10 the x and y directions. Over 4 hours,
centered on meridian passage, the
rms estimation is 0.015 λ/D for both Figure 4 shows the campaign progress in
the x and y directions. Bottom: The data hours collected per night. The maxi-
0.05 relative positions of the field selector mum possible time for which a Centauri
recorded in the same night. The varia-
could be observed at an airmass smaller
tion in the position of the field selector
is due to differential atmospheric than two is about seven hours in a good
0.00 refraction between the AO wave- observing night. Figure 4 shows, how-
–4 –2 0 2
front-sensing channel and the science ever, that there were several consecutive
channel of VISIR. There was an AO
Hour angle (h) nights during the first and last weeks of
interruption at hour angle ~ 2.5–3.
the campaign when either no or only small
amounts of data were recorded. These
(­Carlotti et al., 2012), which does not sup- This method estimates the offsets directly nights suffered from extended periods of
press the overall light intensity, but modi- from the images recorded on the detec- cloud coverage. Even thin high clouds,
fies the light distribution in the focal plane tor. The tests during the commissioning which can be acceptable for observa-
so as to carve out a dark high-contrast phase allowed us to optimise the QACITS tions in the near-infrared, are very detri-
region at the relevant angular separation. algorithm parameters and the observing mental for thermal infrared observations,
Both concepts work well and improve strategy. It was shown that background because they lead to very high fluctua-
the contrast by a factor of between 50 and residuals after chopping have to be sub- tions in throughput and sky background.
100. What tipped the balance towards tracted from the images analysed by
the AGPM as the choice for the NEAR QACITS. After tuning, ­Q ACITS was able
campaign was the higher throughput, to automatically centre the star on the Solid N2 on the coronagraph
resulting in a moderately improved sensi- AGPM and keep it there with an accuracy
tivity overall and, more importantly, the of 0.015 λ/D rms, almost one-hundredth There were, of course, a number of
suppression of the high-intensity stellar of a resolution ­element (Figure 3). smaller and larger problems during the
image, thus avoiding detector electronics long campaign and lots of stories to
“ghosts”. tell. Here is a particularly interesting one,
One hundred hours of observations which concerns one of the unknown
As with all small inner working angle unknowns that we encountered.
coronagraphs, the AGPM performance ESO allocated 20 observing nights
is sensitive to small offsets of the star for the NEAR campaign between 23 May During the first few nights of the cam-
behind the coronagraph (for example, and 11 June 2019 to observe the paign, we noticed that the contrast
slow drifts). In order to actively control the a Centauri system. Even though the ­provided by the coronagraph was less
centring of a Centauri behind the AGPM observing efficiency of NEAR is very effective than during commissioning, with
during the observation, we implemented high, with very small overheads for tele- a continuing slow degradation every other
an algorithm called “Quadrant Analysis scope sky offsets and data transfer night. While we were expecting a suppres-
of Coronagraphic Images for Tip-tilt (well below 10%), the campaign struggled sion of the central point spread function
Sensing” (QACITS; Huby et al., 2015). to collect the 100 hours of data desired. (PSF) by a factor of about 120, we started

The Messenger 178 – Quarter 4 | 2019 7

Telescopes and Instrumentation Kasper M. et al., NEAR: First Results from the Search for Low-Mass P
­ lanets in a Cen

Separation (λ/D) Figure 5. Left: The deepest ever

0 5 10 15 20 25 30 view of the habitable zone (indi-
10 –2
cated by the dashed circle) around
a Centauri A; the 76-hour image
obtained during the NEAR cam-
paign — ~ 6 × 6 arcseconds.

Sensitivity (Jy, 5σ)

Right: Sensitivity and contrast
­e stimated from the deep image as

10 –3 10 –5 a function of radial distance to the

Affected strongly
by ADI residuals

10 –4 10 –6

0 2 4 6 8 10
Separation (arcseconds)

with a factor of 80, which degraded to out the mini-warmup and cool-down in to the position between the two off-axis
only a factor of 40 about a week into the the morning and be back in business in PSFs and binned the surviving 76 hours
campaign. Somewhat frustratingly, none less than 10 hours, sufficiently quick to of data to 1-minute time resolution, which
of the typical external effects that degrade be ready for the following observing is short enough to avoid any noticeable
coronagraph efficiency (for example, Lyot night. And it was a success! The corona- smearing of the images because of field
stop misalignments or optical aberra- graphic rejection recovered with each rotation. This procedure compressed the
tions) could explain the shape of the mini-warmup, and starting from 1 June full campaign into ~ 4600 frames or 3 Gb.
residual image that we observed. It really we repeated this procedure approximately
looked like an intrinsic degradation of the every three days. A relatively simple high-contrast imaging
coronagraphic mask itself. analysis can help evaluate the detection
limits reached during the campaign. By
Could air, entering the cryostat through The data and preliminary results sorting all the frames according to their
a known tiny leak, freeze out on the parallactic angle, we run a PSF calibration
20-Kelvin cold coronagraphic mask and The campaign data were taken at a procedure based on principal component
produce the loss of contrast? A back-of- VISIR/NEAR detector frame rate of 166 Hz, analysis using all the frames, i.e., pro-
the-envelope estimate showed that such i.e., the detector integration time (DIT) cessing the campaign as a whole rather
a leak could indeed build up an ice layer was 6 ms. Chopping ran at 8.33 Hz for than night-by-night. The calibrated images
of a few microns thickness every day. most of the campaign, and each chop- are then combined using noise-weighted
With the refractive index of solid nitrogen, ping half-cycle thus lasted 60 ms. During averages in order to properly take into
the main constituent of air, ice partly this time span, 48 ms or 8 DITs were account the rather large variations in the
entering the grooves of the AGPM coro- averaged into a single frame, and 12 ms sky background.
nagraphic mask could change the opti- or 2 DITs were skipped for the transition
cal depth of the grooves sufficiently to of the DSM between the two chopping Figure 5 shows the result of this simple
degrade the performance. positions. Each 30-second data file data reduction and the contrast sensitivity
­consists of 500 half-cycle frames, and achieved. The 5σ background-limited
So, how were we to test this theory, and the 100 hours of data add up to 6 million sensitivity far away from the star is of the
even more importantly, fix it during the frames or 6 Tb. order of 100 μJy, which is consistent
campaign as it was running? Solid nitro- with our initial goals. At ~ 1.1-arcsecond
gen starts to sublimate at a sufficiently Before entering advanced high-contrast separation, i.e., at the angular size of
high rate to de-ice the coronagraph mask imaging data reduction, some pre-pro- the habitable zone around α Centauri A,
at temperatures that are only moderately cessing was necessary to remove bad the sensitivity is reduced to about 250 μJy
higher than the nominal 20 Kelvin. It frames and reduce the data volume to a mostly by the central glow of the AGPM.
turned out that the temperature after the more manageable size. We removed This does not yet mean that a point
first stage of the instrument warmup, frames with extremely high or variable source can readily be detected at this
­lasting just a few hours, is 35–40 Kelvin. background produced, for example, by level, but first estimates using a fake
Tricking the PLC-controlled system into thin clouds or low encircled energy for injected source show that a planet of
stopping the warmup sequence after the the off-axis stars during ineffective AO ~ 350 μJy brightness corresponding to a
first stage and going into cooling again correction, and frames with low corona- temperate Neptune could indeed be seen.
was risky (a glitch could have resulted in graphic suppression through bad cen-
a full warmup which would have taken tring of the PSF on the coronagraph No planet candidate of the size of Neptune
out VISIR for several days), but it paid mask. Finally, we cropped the images to or larger was found in the data so far.
off. A procedure was developed to carry 400 × 400 pixels, carefully centred them While we were obviously hoping for a

8 The Messenger 178 – Quarter 4 | 2019

detection, the result can also be seen as While no planet candidates have been telescopes, Gemini South and Magellan,
good news for the existence of rocky found so far, NEAR is already a very true Earth analogues could soon be
planets, which may therefore still exist in ­successful collaboration between ESO, discovered.
the habitable zone of a Centauri in a sta- the Breakthrough Initiatives1 and many
ble orbit. There is also a roughly 35% partners in the exoplanet and mid-infrared
chance that an existing planet would astronomy communities. Several key Acknowledgements
have been hidden by the star as the technologies for mid-infrared high-contrast The NEAR experiment greatly benefited, and still
result of an unfavourable projected orbital imaging were successfully tested on-sky, benefits, from the exchange with the exoplanet
position during our single-epoch obser- and many important assumptions were and mid-infrared scientific community on both sides
vation. In addition, the image in Figure 5 validated — for example, the scaling of of the Atlantic. We would like to thank Derek Ives
for access to the Infrared Lab at ESO, Paranal’s
shows some straight lines connecting the the achieved signal-to-noise ratio with the mechanical workshop for the excellent support dur-
coronagraphic centre field with the off- square-root of the observing time. ing the integration on-site, and Rus Belikov, Eduardo
axis stellar image to the lower left. These Bendek, Anna Boehle, Bernhard Brandl, Christian
streaks appear because of a small per- All raw data obtained during the 100-hour Marois, Mike Meyer and Kevin Wagner for very help-
ful discussions and their interest in the data analysis.
sistence in the detector, i.e., the pixels α Centauri campaign are publicly available, Many thanks go also to our industrial partners KT
remember the stars being dragged over and a condensed easy-to-use 3 Gb Optics, Optoline and the Infrared Multilayer Labora-
the detector during the chopping transi- package of all the good frames is availa- tory of the University of Reading (now Oxford), for
tion. This feature is difficult to model and ble on request 3. The on-sky contrast at their R&D spirit and their willingness to stay with us
during the rapid development of the experiment.
may hide another 5–10% of the possible 3 λ/D and the N-band sensitivity are
planet orbits. unprecedented in ground-based astron-
omy by a large margin — more than one References
order of magnitude. The sensitivity limits
Anglada-Escudé, G. et al. 2016, Nature, 536, 437
Beyond NEAR are well understood and could be Carlotti, A. et al. 2012, Proc. SPIE, 8442, 844254
improved further by a factor 2–2.5, mainly Huby, E. et al. 2015, A&A, 584, A74
Preliminary results of the NEAR commis- by removing the AGPM glow by introduc- Ives, D. et al. 2014, Proc. SPIE, 9154, 91541J
sioning and experiment have triggered ing a small optical relay incorporating a Kasper, M. et al. 2017, The Messenger, 169, 16
Lagage, P. O. et al. 2004, The Messenger, 117, 12
substantial interest within the community cold pupil stop in front of the AGPM. But Mawet, D. et al. 2005, ApJ, 633, 1191
in this facility, and also for other astro- this is still not the limit for mid-infrared N’Diaye, M. et al. 2014, Proc. SPIE, 9148, 91485H
nomical observations. ESO therefore observations from the ground. A novel
issued a call for Science Demonstration lower-noise detector technology is
proposals, which received a lot of atten- emerging, which promises to double the
tion and resulted in 26 proposals being sensitivity once more. These next-gener- 1
 reakthrough Initiatives webpage: http://break-
submitted for NEAR observing time. Two ation detectors would allow the VLT to
periods of Science Demonstration were probe the rocky planet regime in the hab- SPARTA:
allocated in September and December itable zone around a Centauri. When 3
Data can be requested via e-mail from Prashant
2019 to conduct roughly half of the pro- combined with similar instruments at the Pathak ( or Markus Kasper
posed programmes. other southern hemisphere 8-metre-class (
ESO/NEAR Collaboration

The NEAR experiment being mounted on the

Cassegrain focus of the VLT’s UT4 (Yepun).

The Messenger 178 – Quarter 4 | 2019 9

Telescopes and Instrumentation DOI: 10.18727/0722-6691/5164

Report on Status of ESO Public Surveys and

Current Activities

Magda Arnaboldi 1 ESO Public Surveys: overview of sis, including the timeline for the delivery
Nausicaa Delmotte 1 engagement rules and status of science data products over the entire
Dimitri Gadotti 1 duration of the survey project. The
Michael Hilker 1 By design, the ESO Public Surveys cover approval of the survey management plan
Gaitee Hussain 1 a variety of research areas, from the is further confirmed by the agreement
Laura Mascetti 2 detection of planets via micro-lensing, signed between ESO’s Director General
Alberto Micol 1 through stellar variability and evolution, the (DG) and the Principal Investigator (PI) of
Monika Petr-Gotzens 1 Milky Way and Local Group galaxies, to each survey.
Marina Rejkuba 1 extragalactic astronomy, galaxy evolution,
Jörg Retzlaff 1 the high-redshift Universe and cosmol- The agreement between the DG and the
Chiara Spiniello 1, 3 ogy. Differently from Large Programmes, PI specifies the milestones for the data
Bruno Leibundgut 1 these projects are planned to span more releases and their content and responsi-
Martino Romaniello 1 than four semesters and last for many bility for the scientific quality and accu-
years. For example, the latest call for the racy of the data products, which is to be
Cycle 2 survey projects for the Visible warranted by the Public Survey team
ESO and Infrared Survey Telescope for Astron- under the leadership of the PI. The agree-
Terma GmbH, Darmstadt, Germany omy (VISTA) required them to span a time ment states that a final release which
Astronomical Observatory of interval of more than three years. These includes the reprocessing of the entire
­Capodimonte, Naples, Italy a survey projects all have a legacy value for data set is expected upon completion of
the community at large in addition to the data acquisition for each survey. This final
science goals identified by the proposing data release should take place within
This report on the status of the ESO teams. one year of completion of the data acqui-
Public Surveys includes a brief overview sition for any survey. The PSP was set up
of their legacy value and scientific to periodically review the progress of the
impact. Their legacy is ensured by their ESO science policies for Public Surveys surveys and to assess compliance with
homogeneity, sensitivity, large sky the specification of the survey products.
­coverage in multiple filters, large num- The selection of ESO Public Surveys is In May 2019, a PSP review took place to
ber of targets, wavelength coverage a two-step process which starts with the evaluate the scientific impact of the active
and spectral resolution, which make submission of letters of intent. On the Public Surveys.
them useful for the community at large, basis of these letters, the Public Survey
extending beyond the scientific goals Panel (PSP) formulates a coherent, well-
identified by the survey teams. In May balanced scientific programme that takes Operations for ESO Public Surveys
2019, as almost all first-generation into account any synergies among teams
imaging and spectroscopic surveys in the community and the international The ESO Public Survey observations —
completed their observations and second- survey projects. The PSP then provides whether in service mode or visitor mode
generation imaging surveys got well recommendations to ESO including a list — are carried out according to the pro-
underway, the Public Survey Panel of the teams that should be invited to cess defined by the ESO Data Flow Sys-
reviewed the scientific impact of these submit full proposals on the basis of the tem. The raw data acquired for ESO
projects. The review was based on a ranking of the descriptions of their sci- ­Public Surveys are immediately public.
quantitative assessment of the number ence projects as provided in the letters of Once the Public Survey teams have car-
of refereed publications from the survey intent. In so doing, the PSP fosters active ried out data reduction to remove instru-
teams and archive users. It included collaborations within the community by mental signatures, calibrate the data and
the number of citations, the number of asking independent teams to join, encour- complete the measurements defined by
data releases and statistics on access aging them to optimise science goals their scientific goals, ESO assists the
to archive data by the user community. and observing strategies, and sharing ­survey teams to define and package their
The ESO Users Committee also dis- resources. data products in a manner consistent
cussed the availability and usage of with the ESO Science Archive and Virtual
ESO Public Survey data by the commu- Once the proposals have been recom- Observatory standards and in agreement
nity during their yearly meeting in April mended for approval by the PSP and with the specifications in the survey
2019. We describe the status of these the Observing Programmes Committee, ­management plans. The goal is to inte-
projects with respect to their observing data acquisition for each ESO Public grate science data products from the
plans, highlight the most recent data ­Survey starts. This involves the review Public Surveys into the ESO Archive,
releases and provide links to the result- and assessment of each survey manage- together with the entire archive content
ing science data products. ment plan by the ESO Survey Team. from the La Silla Paranal Observatory.
The survey management plan is an This is done via the Phase 3 process,
essential tool for the survey team, as well which is an audit process that certifies
as for operations at ESO; it details the the integrity, consistency and data quality
data acquisition plan, and the allocated of the products available from the ESO
resources for data processing and analy- Archive and ensures a homogeneous

10 The Messenger 178 – Quarter 4 | 2019

user experience once the data are pub- 120%
lished through the Archive.

Percentage of completion (OB hours)


ESO Public Survey status

A total of twenty Public Surveys1 have
been carried out by consortia in the com- 60%
munity and are actively supported by
ESO. The majority have completed data
acquisition using ESO facilities and are 40%
in the process of publishing science data
products via the ESO archive. 20%
ESO Public Surveys were launched in VPHAS+
2005 with an initial call for the optical
1 8
12 2
1 4
1 8
13 2
1 4
1 8
1 2
1 4
1 8
15 2
1 4
1 8
16 2
1 4
1 8
17 2
1 4
18 2
1 4
1 8
19 2
imaging surveys at the VLT Survey Tele-
20 1-1

20 2-1

20 3-1

20 4-1

20 5-1

20 17-1
20 6-1

20 8-1
20 - 0
20 1-0

20 2-0

20 - 0
20 3 - 0

20 4 - 0
20 4 - 0

20 - 0
20 5 - 0

20 - 0

20 - 0

20 - 0
20 6 - 0

20 7-0

20 8 - 0


scope (VST; Capaccioli & Schipani, 2011),


followed by a call for the near-infrared Date

surveys (Cycle 1) in 2007 (Arnaboldi et al.,
2007) at VISTA (Sutherland et al., 2015). 120%
Percentage of completion (OB hours)

Once the imaging surveys were under VHS

way, ESO opened a first call for Public 100% UltraVISTA
Spectroscopic Surveys in 2011, followed VIDEO
by a second call for the VIMOS Public 80% VVV
Spectroscopic Surveys in 2015. The call VMC
for Cycle 2 VISTA imaging Public Surveys 60%
was opened in 2015 and the selected VVVx
surveys began in April 2017 (Arnaboldi et G-CAV
al., 2017). Four of the seven Cycle 2 VISTA
surveys exploit the time domain: for SHARKS
example, following up exotic transients 20% UltraVISTA-New
like the optical-near-infrared echo of grav- VISIONS
itational wave (GW) events (VinRouge); 0% VinRouge
studying the 3D shape of the Milky Way
10 0
1 2
11 0
1 2
12 0
1 2
13 0
1 2
14 0
1 2
15 0
1 2
16 0
1 2
17 0
1 2
18 0
1 2
19 0
19 2
20 7-1
20 0-1

20 2-1
20 9-1

20 1-1

20 3-1

20 4-1

20 5-1

20 6-1

20 8-1
20 - 0

20 - 0

20 - 0

20 - 0

20 - 0

20 - 0

20 - 0

20 - 0

20 - 0

20 - 0

bulge (VVVX) via astrometry — to test


stellar evolution models and microlensing, Date

and to obtain proper motion membership
(VVVX, VISIONS); or detecting high-z Figure 1. (Upper) Cumulative curves for the comple- Figure 2. (Lower) Cumulative curves of completion
tion of the VST surveys. The VST ATLAS survey was for Cycle 1 and 2 VISTA Surveys. The cumulative
supernovae (SN) in cosmological deep
extended after the completion of the observing plan curves of completion can reach values over 100% for
fields (VEILS). Two of the seven Cycle 2 to allow further coverage in the u’g’r’-bands (as out- Public Surveys that were compensated for time cor-
surveys, VVVX and the Continuing Ultra- lined in its Survey Management Plan). The VPHAS+ responding to low-quality observations (OBs with a
VISTA, follow up the successful Cycle 1 curve is below 100% because the PI requested that D grade). Public Survey teams can ask for compen-
it be terminated early. sation via reports submitted to the OPC.
surveys, very much in the spirit of other
surveys such as the Sloan Digital Sky
Survey (SDSS). and 2018, except for the VHS South Pole observing time — in hours for the imag-
fields. The data acquisition for the Cycle 2 ing surveys and in nights for the spectro-
In the optical, the VST Public Surveys VISTA imaging surveys is two-thirds scopic surveys.
completed their data acquisition in ­complete. In Figure 2 we show the cumu-
Period 104. The V-ATLAS survey was lative curves of data acquisition for all
granted an extension by the PSP to VISTA surveys. The data acquisition for Scientific impact of ESO Public Surveys
acquire the u’g’r’-band imaging of chosen the four spectroscopic surveys, including
sub-areas. The data acquisition for this the two that were carried out with the A standard reference metric for the
extension is ongoing and completion is VIMOS spectrograph, has also been assessment of scientific impact is given
expected in Period 105. In Figure 1 we completed. by the number of refereed publications
show the cumulative curves of data acqui- from the Public Survey teams. Given the
sition for the VST surveys. The data In Tables 1 and 2 we provide a summary legacy value of these projects and the
acquisition for the Cycle 1 VISTA imaging of the observational parameters for the science data products readily available
surveys was completed between 2015 twenty ESO Public Surveys and the total for download via the ESO Archive, other

The Messenger 178 – Quarter 4 | 2019 11

Telescopes and Instrumentation Arnaboldi M. et al., Report on Status of ESO Public Surveys and Current Activities

VST Survey ID Science Area (square Filters Magnitude limits Total time
degrees) (hours)
KiDS — Kilo-Degree Survey Extragalactic 1350 b u’ g’ r’ i’ 24.1 24.6 24.4 3421 23.4
(de Jong et al., 2013)
ATLAS Wide area/baryon 4700 c u’ g’ r’ i’ z 22.0 22.2 22.2 1585 acoustic oscillations 21.3 20.5
(Shanks et al., 2013)
VPHAS+ — VST Photometric Hα Survey of the Southern Stellar astrophysics 1800 d u’ g’ Hα r’ i’ 21.8 22.5 21.6 1200
Galactic Plane 22.5 21.8 (Drew et al., 2013)

VISTA Cycle 1 Science Area (square Filters Magnitude limits Total time
degrees) (hours)
UltraVISTA Deep high-z 1.7 Deep Y J H Ks 25.7 25.5 25.1 1832 0.73 Ultra deep NB118 24.5 26.7 26.6
(McCracken et al., 2013) 26.1 25.6 26.0
VHS — VISTA Hemisphere Survey Southern sky 17 800 Y J H Ks 21.2 21.1 20.6 4623 20.0
(McMahon et al., 2013)
VIDEO — VISTA Deep Extragalactic Observations Survey Deep high-z 12 Z Y J H Ks 25.7 24.6 24.5 2073 24.0 23.5
(Jarvis et al., 2013)
VVV — VISTA Variables in the Via Lactea Milky Way 560 Z Y J H Ks 21.9 21.1 20.2 2205 18.2 18.1
(Hempel et al., 2014)
VIKING — VISTA Kilo-Degree Infrared Galaxy Survey Extragalactic 1500 Z Y J H Ks 23.1 22.3 22.1 2424 21.5 21.2
(Edge et al., 2013)
VMC — VISTA Magellanic Clouds Survey Resolved star 180 Y J Ks 21.9 21.4 20.3 2047 formation history
(Cioni et al., 2013)

VISTA Cycle 2 Science Area (square Filters Magnitude limits Total time
degrees) (hours)
VINROUGE* — Kilonova counterparts to gravitational wave Kilonova 300 Y J Ks 21.0 21.0 20.1 77
sources counterparts to GW sources
(Tanvir et al., 2017)
Cont. UltraVISTA — Completing the legacy of UltraVISTA High-z 0.75 J H Ks 26.0 25.7 25.3 567
VVVX* — Extending VVV to higher Galactic latitudes Milky Way 1700 J H Ks Ks = 17.5 1631
VEILS* — VISTA Extragalactic Infrared Legacy Survey Galaxy evolution, 9 J Ks J < 23.5 847 AGN, SN Ks < 22.5
G-CAV — Galaxy Clusters At VIRCAM Galaxy clusters 30 Y J Ks 24.5 24 23 440
VISIONS* — VISTA star formation atlas Star formation 550 J H Ks 21.5 20.5 19.5 449 atlas
SHARKS — Southern Herschel-Atlas Regions Near-infrared 300 Ks 22.7 929
Ks-band survey counterparts for radio sources

Table 1. (Upper) VLT Survey Telescope Public Sur- Table 2. (Centre) Cycle 1 VISTA Public Surveys; Table 3. (Lower) Cycle 2 VISTA Public Surveys
veys. These projects began operations in October these projects began operations in April 2010 and began operations in April 2017. The four Cycle 2
2011 and data acquisition is now completed accord- are now all completed but for the VHS subareas VISTA surveys that explore the time domain are
ing to their survey management plans. The total close to the South Galactic Pole. The total number ­indicated by an asterisk in the table. The total
number of completed hours is reported to the 30 of completed hours is reported to the 30 September ­number of completed hours by 30 September 2019
September 2019 date. 2019 date. is shown in the last column.

12 The Messenger 178 – Quarter 4 | 2019

Public Spectroscopic Survey ID Science topic Number of Spectral Total time
and homepage targets/spectra resolution (nights)
Gaia–ESO Milky Way, stellar 200 000 20 000 282.5 populations
(Randich et al., 2013)
PESSTO — Public ESO Spectroscopic Survey of Transient Objects Transient, 150 ~ 2500 384.0 SN progenitors
(Smartt et al., 2013)
VANDELS Physics of galaxies in the 2700 ~ 1500 142.7 early universe CANDELS,
(McLure et al., 2017) UDS & CDFS fields
LEGA-C — Large Early Galaxy Astrophysics Census Dynamics of galaxies 3100 ~ 1500 99.8 at z = 0.6–1.0
(van der Wel et al., 2016)

Table 4. Public Spectroscopic Surveys. PESSTO 700

and Gaia–ESO began operations in 2012 and were
Number of publications/citations

completed in 2017. The surveys using the VIsible

600 Refereed publications
Multi Object Spectrograph (VIMOS), called Citations (since 2010)
­VANDELS and LEGA-C, began operations in 2015
500 Total ref. publications: 848
and were completed in March 2018, before the
172 (20.3%) from archive
decommissioning of the VIMOS spectrograph.
400 84 (10%) are archive + PI

independent archives (for example, the 300

VISTA science archive, Vizier) or the
­Public Survey webpages have also been 200
made available to those in the community
interested in accessing data products for 100
their independent scientific explorations.









The ESO library routinely monitors













­refereed publications, based on data

tr a






acquired from ESO approved observing ESO Public Survey

programmes. This includes papers
­published by PIs/co-investigators (CoIs) Published data releases Figure 3. Histogram of the cumulative number of ref-
ereed publications and citations (divided by 10) for
as well as archive papers. Archive papers
each ESO Public Survey.
come in two flavours: archive only and Because of the extensive amount of time
archive plus PI publications. In archive allocated using ESO facilities, the science
only papers, none of the authors of these policies for ESO Public Surveys entail
refereed publications are listed as PIs the submission and publication of the specific infrastructures. Five out of six
or CoIs of the approved Public Survey ­science data products from these pro- Cycle 1 VISTA surveys and two out of
proposals. In the case of archive plus PI jects into the ESO Archive. The publica- three VST surveys received extensive
publications, science data products from tion process for science data products support from the dedicated data centres,
the ESO Public Surveys are used extends well after the completion of the CASU 3 and WFAU 4. The deep UltraVISTA
together with data owned by a PI or CoI data acquisition. This additional time is and Continuing UltraVISTA surveys relied
of an ESO programme to achieve their used by the Public Survey teams to exe- on dedicated support from CASU,
scientific published results. In the case of cute global calibrations of the entire data ­TERAPIX 5 and CALET 6 centre at the IAP
ESO Public Surveys, the total number of volume and to carry out the relevant in Paris, while the KIDS survey is sup-
refereed publications by teams and measurements required to achieve their ported by Astro-WISE 7. The Cycle 2 VISTA
archive users was 848 by 30 September scientific goals. The ultimate publication surveys have adopted different strategies
2019. Of these refereed publications, 172 of the results of these steps is contained compared to the first generation, with
(20.3%) are archive only and 84 (10%) are in the final catalogue release. All twenty a larger number receiving tailored sup-
archive plus PI since 2010 (from ESO tel- ESO Public Surveys are currently involved port to their data processing from their
bib 2). The total number of citations from in the publication of their science data respective science institutes.
ESO Public Survey refereed publications products via the ESO Archive.
is 26 266. In Figure 3 we provide the his- For the Public Spectroscopic Surveys,
togram of the cumulative number of refer- The Public Survey teams adopted a Gaia-ESO, PESSTO, LEGA-C and
eed publications and citations per survey range of strategies to deal with the data ­VANDELS, the teams built their data
project. volumes from their respective surveys. reduction infrastructure based on previous
Some rely on the support of data centres experience they had acquired through
while others have developed their own managing large programmes at ESO and

The Messenger 178 – Quarter 4 | 2019 13

Telescopes and Instrumentation Arnaboldi M. et al., Report on Status of ESO Public Surveys and Current Activities

scientific networks (for example, of the Small Magellanic Cloud. All data of files downloaded by the community for
PESSTO–WISeREP). releases were promptly advertised via the each ESO Public Survey. The lower chart
Archive/Phase 3 web pages, followed up shows the numbers of catalogues, the
All survey teams have successfully with specific announcements on the ESO numbers of distinct users and the num-
­published several data releases for some science page, the Science News­letter 9 bers of queries carried out using the ESO
of their science data through the Phase 3 and the ESO archive community forum10. catalogue query interface 12 to access
process (Arnaboldi et al., 2014); an over- ESO Public Survey catalogues. On aver-
view of these releases is available via The most recent data releases join a age, users of the ESO catalogue query
this webpage8. Since January 2019, the large number of data collections from the interface carry out at least 21 independ-
total volume of science data products ESO Public Surveys that can be browsed ent queries to access catalogue records.
released from the ESO Public Surveys using the Archive Science Portal11. The
amounts to 27.4 Tb, including ancillary science data products from the ESO An enhanced archive capability allowing
files. The data releases published this Public Surveys amount to a total volume programmatic access13 results in anony-
year include: the fourth data release of of 68.6 Tb (nearly 8.5 × 105 files) which mous exploration and retrieval of cata-
KIDS (> 1000 square degrees) and are currently accessible via the ESO logue records (and other products) via
­UltraVISTA (deep stacked images of the Archive. The science data products that Virtual Observatory tools, for example,
COSMOS field from observations can be actively queried and downloaded Aladin and Topcat. This new service
acquired between December 2009 and amount to nearly 320 000 catalogue files, allows users to repeat queries in an auto-
June 2016); the proper motion of selected half a million astrometrically and photo- mated fashion, in order to perform more
stars in the Milky Way disc and bulge metrically calibrated images, and 56 000 complex queries by combining data from
from the VVV near-infrared Astrometric 1D extracted spectra. In Figure 4 we different surveys or other content of the
Catalogue (VIRAC); accurate PSF-fitting show a collection of on-sky footprints of ESO Science Archive, thereby enhancing
photometry of the 300 square degrees the data releases published during the the scientific use of the catalogue content
around the Galactic centre; and the fifth last year by the ESO Public Survey teams. of the ESO Archive. One interesting
data release of VMC with full coverage ­statistic is the number of distinct users —
1583 users from 77 different countries —
Data download statistics who have downloaded ESO Public
Figure 4. Montage of the footprints of the data
releases from the ESO Public Surveys published by
Survey science products published via
2019, as shown on the ESO Archive Science Portal In Figure 5 we show the cumulative curves the ESO Archive. To place this in context,
interactive interface.   of the data volume (Gb) and the number the fraction of distinct users who access




14 The Messenger 178 – Quarter 4 | 2019

18 000 250 000

16 000

14 000 200 000

12 000
Data volume (Gb)

Number of files
150 000
10 000

100 000

50 000

0 0






data products from ESO Public Surveys Figure 5. Charts show- 2500
ing the cumulative data
is 46% of the total number of users
volume (upper left) and
(3457) who have downloaded science
Number of catalogues/users/queries

numbers of files down-

data products from the ESO Archive. loaded (upper right)
for each ESO Public
Survey. The lower chart
shows the cumulative 1500
How ESO promotes Public Survey curves for the number of
science catalogues (× 10), the
number of distinct
users, and number of
ESO promotes science results from
queries (× 0.1) related to
­Public Surveys via science/photo press the ESO Public Surveys
releases and community workshops. downloaded using 500
The most recent photo release of the new the ESO catalogue
query interface.
image of the Large Magellanic Cloud14
celebrated the 500th anniversary of its 0

O 3

Ju 4

Ju 5

Ju 6

Ju 7

Se 8
M 3

O 4




first being seen by Europeans (during 18













the voyage of the explorer Ferdinand




Number of queries/10 Distinct users Number of catalogs (× 10)

Magellan). Among the science press
releases, we particularly note those on
the 3D structure of the bulge of the Milky
Way15, the constraints on the clumpiness Community awareness of ESO Public ESO Archive in the vast majority of cases.
of the dark matter distribution in the Surveys: feedback from the User The UC poll participants who had used
­Universe16, and the first light from a gravi- ­Community poll science data products from Public Sur-
tational wave source17. veys for their own science reported that
In April 2019 the ESO Public Surveys they had published between one and
Three community workshops were were identified as the special topic for the three peer-reviewed papers based on
organised by ESO in 2012, 2015 and User Community (UC) poll. The results of these data. Finally, most users who had
2019 to support survey science and this UC poll showed that more than 60% participated in the surveys or used Public
operations at ESO. The most recent was of the ESO users who responded were Survey data provided a positive assess-
the 4MOST community workshop (see aware of the ESO Public Surveys. About ment of their utility, scheduling, rate of
also The Messenger 17518, and Liske & 25% of UC poll participants (excluding all progress/publication, and the effective-
Mainieri, 2019). A workshop focused on Survey PIs/CoIs) had used archive data ness of communication with ESO. Regard-
the next-generation galaxy evolution from Public Surveys to complement other ing the services provided by ESO, i.e.,
­surveys is currently being organised in datasets. This was the most frequent archive interfaces and Phase 3, the vast
Perth, Australia in February 2020. This usage of these data products according majority of the users were “satisfied” or
second Australia-ESO conference will to the results of the poll. “very satisfied”, with the ESO Archive
discuss the future coordination of these being the preferred site for data retrieval.
surveys with multi-wavelength facilities in The science data products from ESO The release description published together
the southern hemisphere. Public Surveys were retrieved from the with an active release was considered

The Messenger 178 – Quarter 4 | 2019 15

Telescopes and Instrumentation Arnaboldi M. et al., Report on Status of ESO Public Surveys and Current Activities

the most reliable/useful documentation to announced 2019 a call for letters of intent Drew, J. E. et al. 2013, The Messenger, 154, 41
Edge, A. et al. 2013, The Messenger, 154, 32
support the scientific use of the available for 4MOST Community Surveys on
Hempel, M. et al. 2014, The Messenger, 155, 29
data products. 26 November 2019. The community sur- Jarvis, M. J. et al. 2013, The Messenger, 154, 26
veys will access up to 30% of the availa- Liske, J. & Mainieri, V. 2019, The Messenger, 177, 61
ble time on the 4MOST spectrograph McCracken, H. J. et al. 2013, The Messenger, 154, 29
McLure, R. et al. 2017, The Messenger, 167, 31
The Public Survey Panel review on VISTA over a period of five years.
McMahon, R. et al. 2013, The Messenger, 154, 35
These projects will complement the GTO Randich, S. et al. 2013, The Messenger, 154, 47
Membership and evolution of the Public surveys that were presented during the Shanks, T. et al. 2013, The Messenger, 154, 38
Survey Panel 4MOST workshop at ESO in May and Smartt, S. et al. 2013, The Messenger, 154, 50
Sutherland, W. et al. 2015, A&A, 575, 27
are described in the special issue of the
Tanvir, N. et al. 2017, ApJ, 848, 27
In 2005, the Public Survey Panel (PSP) Messenger in March 2019 7. van der Wel, A. et al. 2016, The Messenger, 164, 36
was set up as a subpanel to the Observ-
ing Programmes Committee, its role Large surveys are considered a key tool
being later extended to include monitor- in observational astronomy because
ing of the progress of the Public Survey they allow explorations that go beyond 1
ESO Public Surveys Project webpage:
projects. Four chairs organised the work individual targeted observations. They are
of the panel: Duccio Macchetto, Simon characterised by large investments that sciencePublicSurveys.html
ESO Telescope Bibliography telbib:
White, Danny Lennon, and Miguel Mas comprise dedicated telescopes and
Hesse (the current chair). From 2005 to instruments, a wide community of astron- 3
The Cambridge Astronomy Survey Unit:
2012, ESO set up three separate panels: omers involved in the science projects
the VISTA, VST imaging, and Spectro- and extended networks for the data distri- 4
The Wide Field Astronomy Unit:
scopic Public Survey Panels; the VISTA bution and analysis. The scientific success 5
and VST Panels were later merged. Since of such survey projects includes the leg- 6
2018 there has been only one panel for acy value of science products that are 7
all ESO Public Surveys, whether imaging made available through the archives for 8
ESO Phase 3 Data Releases:
or spectroscopy. In the near future, the further scientific exploration by the com- 9
ESO Science Newsletter:
PSP will advise ESO on the selection of munity. ESO Public Surveys are an exam- publications/newsletter/
4MOST surveys (see de Jong et al., 2019). ple of an effective implementation of this 10
ESO Archive Community Forum:
strategy, with the goal of supporting the
ESO Archive Science Portal:
scientific advancement of its community.
The PSP review in May 2019 12
ESO catalogue query interface:
As most of the Public Surveys completed Acknowledgements 13
ESO Archive Programmatic Access webpage:
their data acquisition in 2019, the goal The authors would like to thank colleagues at the 14
VISTA image of LMC:
and objectives of the PSP review held in La Silla Paranal Observatory for their work, and sup-
May at the ESO headquarters were to port by the science operations of the ESO Public 15
VISTA image of Milky Way bulge:
assess the scientific success of the Public Surveys. They wish to acknowledge the colleagues
from the ESO Directorate of Engineering who sup- 16
VISTA image from KIDS survey:
Surveys using criteria that included pub- ported the development of the tools required for car-
lished results in refereed journals, the rying out Phase 1, Phase 2 and Phase 3 operations 17
First Light from Gravitational Wave Source:
progress of the analysis, potential scien- for ESO Public Surveys and the Archive science
tific extensions, complementarity with interfaces. The authors thank the ESO library team 18
Messenger issue dedicated to 4MOST GTO
for their careful monitoring of the refereed publica- ­surveys:
other telescopes, activities to promote tions. The authors wish to thank the PIs of the Public messenger/archive/no.175-mar19/messenger-
the surveys, and the use of survey data Surveys and their collaborators, including the data no175.pdf
products for independent projects in the centres at CASU3, WFAU4, Astro-WISE14 (Astro­
community. The PIs of the twenty active nomical Wide-field Imaging System for Europe) and
TERAPIX5, for their hard work and support of the Notes
Public Surveys were invited to the review ESO mission. Finally, ESO wish to thank the PSP
and asked to address the above criteria chairs and the colleagues in the community who a
 strofit2 Marie Skłodowska-Curie Fellow
during their presentations. All PIs but one served as members in the different PSPs for their b
T he survey area was reduced from the original
attended the review and contributed to efforts and their support with the definition of how approved coverage to coincide with the VIKING
to execute the survey programmes. footprint.
a lively and constructive discussion. The c
T he survey area was increased to 4700 square
PSP report was presented to the Scien- degrees after approval by the PSP; data-taking is
tific Technical Committee at its meeting in References currently active.
October 2019. d
T he survey area was reduced from that approved
Arnaboldi, M. et al. 2007, The Messenger, 127, 28 by the PSP following a request by the PI to end the
Arnaboldi, M. et al. 2014, The Messenger, 156, 24 survey early.
Arnaboldi, M. et al. 2017, The Messenger, 168, 15
Outlook and Conclusions Capaccioli, M. & Schipani, P. 2011, The Messenger,
146, 2
Cioni, M.-R. et al. 2013, The Messenger, 154, 23
While work on the scientific analysis con- de Jong, R. 2019, The Messenger, 175, 3
tinues for the twenty Public Surveys, ESO de Jong, R. et al. 2013, The Messenger, 154, 44

16 The Messenger 178 – Quarter 4 | 2019

Telescopes and Instrumentation DOI: 10.18727/0722-6691/5165

MUSE Spectral Library

Valentin D. Ivanov 1 sparse coverage of the parameter space

Lodovico Coccato 1 (Pickles, 1998; Le Borgne et al., 2003; 0
Mark J. Neeser 1 Yan et al., 2018).
Fernando Selman 1
Alessandro Pizzella  2, 3 Spectral datasets that are available
Elena Dalla Bontà  2, 3 include the Elodie library (Soubiran et 2
Enrico M. Corsini  2, 3 al., 1998; Prugniel & Soubiran, 2001;

log g
Lorenzo Morelli  4 Le Borgne et al., 2004) and the X-shooter
Spectral Library (XSL; Chen et al., 2014).
The latter showcases the problems that
ESO increasing resolution and multi-order 4
Dipartimento di Fisica e Astronomia cross-dispersed spectrographs bring;
“G. Galilei”, Università di Padova, Italy synthetic broadband optical (UBV) colours
INAF–Osservatorio Astronomico di show poor agreement with observed
Padova, Italy ­colours from the Bright Star Catalogue 4.5 4 3.5
Instituto de Astronomía y Ciencias (on average at ~ 7%; see Table 5 and Fig- log Teff
­Planetarias, Universidad de Atacama, ure 26 in Chen et al., 2014). The differ-
Copiapó, Chile ences are likely related to pulsating varia-
ble stars observed at different phases. 6
Slit losses are another issue; for many 4

Empirical stellar spectral libraries have stars these are caused by the attenuation
applications in both extragalactic and of flux, or other losses inherent to slit-
stellar studies. We have assembled the based spectrographs. 0
MUSE Spectral Library (MSL), consist- O B A F G K M C/S
ing of 35 high-quality spectra of stars We embarked on a project to build an Spectral type
covering the Hertzsprung–Russell dia- empirical spectral library without slit
gram, and verified the continuum shape losses using the MUSE (Multi-Unit Spec- Figure 1. Properties of the MSL stars. Top: surface
gravity log g vs. effective temperature Teff for stars with
of our spectra with synthetic broadband troscopic Explorer; Bacon et al., 2010)
[Fe/H] ≤ −0.5 dex (crosses), −0.5 < [Fe/H] < 0.0 dex
colours. We also report indices from IFU, with the goal of spanning all of the (open circles), and [Fe/H] ≥ 0.0 dex (filled circles).
the Lick system, derived from the new major sequences on the Hertzsprung– Bottom: distribution of the stars by spectral type.
observations. Our data demonstrate Russell diagram and serving as a bench-
that integral field units (IFUs) are excel- mark for the shapes of other theoretical
lent tools for building spectral libraries and empirical spectra. Our final products Data reduction was performed with the
with reliable continuum shapes that can are spectra suitable for galactic model- ESO MUSE pipeline (v. 2.6) within the
be used as templates for extragalactic ling, stellar classification and other appli- ESOReflex 3 environment (Freudling et
studies. cations. Here we report on our first sam- al., 2013). The 1D spectra were extracted
ple of 35 MSL spectra. using a circular aperture with a radius of
6 arcseconds. This number was selected
Introduction and sample Our initial sample numbered 33 XSL stars1. after experiments with different aperture
In addition, HD 193256 and HD 193281B sizes, to guarantee that “aperture” losses
Empirical stellar spectral libraries are a serendipitously fell inside the MUSE field led to less than a 1% change in the over-
universal tool in modern astronomy, with of view. The full sample is described in all slope of the spectra from the blue to
applications in both extragalactic and Table 1 of Ivanov et al. (2019). the red.
galactic stellar studies. They can have
multiple uses: to match and remove Three stars were treated differently,
­continua to reveal weak emission lines; Observations and data reduction ­without major loss of continuum fidelity.
as templates to measure stellar kinemat- For the asymptotic giant branch star
ics in galaxies; and to measure stellar The spectra were obtained with MUSE [B86] 133 we reduced the extraction
parameters such as effective tempera- at the European Southern Observatory aperture radius to 4 arcseconds to avoid
tures and surface gravities. Theoretical (ESO) Very Large Telescope, Unit Tele- contamination from nearby sources. For
stellar models can have significant weak- scope 4 (Yepun), on Cerro Paranal, Chile. HD 193256 the aperture had a radius
nesses; for example, Sansom et al. (2013) Table A.1 in Ivanov et al. (2019) gives the of 4.6 arcseconds, and the sky annulus
found discrepancies in Balmer lines and observing log. We obtained six exposures had an inner radius of 4.6 arcseconds
the incomplete treatment of molecules for each target, except for HD 204155 and a width of 2 arcseconds because the
(also shown by Castelli, Gratton & which was observed 12 times. We placed star was close to the edge of the MUSE
Kurucz, 1997). This occasionally leads to the science targets at the same spaxels field of view. HD 193281 is a binary with a
poorly predicted broad-band colours. At as the spectrophotometric standards to separation of ~ 3.8 arcseconds. We dis-
the same time, the typical empirical minimise any residual systematics from entangled the two spectra as described
libraries suffer from low resolution and/or the instrument. in Ivanov et al. (2019).

The Messenger 178 – Quarter 4 | 2019 17

Telescopes and Instrumentation Ivanov, V. D. et al., MUSE Spectral Library


HD 057060 07e

10 HD 096446 B2III

Fλ (arbitrary units + const)

Fλ (arbitrary units + const)

HD 174966 A3
HD 167278 F2
HD 163810 G3V 6

HD 167278 K3V

(B86) 133 M4

0 HD 067507 CNv 0
6000 8000 10 4 4850 4900 4950 6600 6700 8500 8600 8700
Wavelength (Å) Wavelength (Å) Wavelength (Å) Wavelength (Å)

Figure 2. Comparison of a subset of our MSL spectra fitted second-order polynomials to the University through grants DOR1715817/17,
(black) with the XSL spectra (red; boxcar smoothed DOR1885254/18, DOR1935272/19, and
ratios and extrapolated them over the full
over 8 pixels). The spectra are normalised to unity BIRD164402/16.
between the two vertical dotted lines shown on the wavelength range covered by the XSL
left, and shifted vertically for display purposes. Left: library to demonstrate that, if these
entire MUSE spectral range; right: zoom around the trends hold, the overall peak-to-peak flux References
Hβ, Hα, and Ca triplet features (left to right). No
differences can easily reach ~ 20%,
radial velocity correction is applied. Bacon, R. et al. 2010, Proc. SPIE, 7735, 773508
meaning that the overall continua of the Castelli, F., Gratton, R. G. & Kurucz, R. L. 1997,
cross-dispersed spectra is somewhat A&A, 318, 841
The final MSL spectra have signal-to- ill-defined. Chen, Y.-P. et al. 2014, A&A, 565, A117
Freudling, W. et al. 2013, A&A, 559, A96
noise ratios S/N > 70–200 and are availa-
Ivanov, V. D. et al. 2019, A&A, 629, 100
ble via the ESO MUSE webpage 2 or via Finally, we calculated synthetic Sloan Le Borgne, J.-F. et al. 2003, A&A, 402, 433
CDS/VizieR 3. The Lick indices (Worthey Digital Sky Survey (SDSS) colours from Le Borgne, D. et al. 2004, A&A, 425, 881
et al., 1994) that fall within the wavelength both MSL and XSL spectra (Figure 5 in Pickles, A. J. 1998, PASP, 110, 863
Prugniel, P. & Soubiran, C. 2001, A&A, 369, 1048
range covered by MUSE were measured Ivanov et al., 2019). The MUSE sequences
Sansom, A. E. et al. 2013, MNRAS, 435, 952
in the new MSL spectra (Table C.1 in are slightly tighter than the XSL ones, con- Soubiran, C. et al. 1998, A&AS, 133, 221
­Ivanov et al., 2019). firming that the IFU MUSE spectra have Yan, R. et al. 2019, ApJ, 883, 175
more reliable shapes. This is expected in Worthey, G. et al. 1994, ApJS, 94, 687
light of the slit losses and the imperfect
Analysis and discussion order stitching of the XSL spectra. Fur- Links
thermore, X-shooter has three arms and
We demonstrate excellent agreement is in effect three different instruments; 1
 he XSL library:
T he MUSE spectral library at the ESO MUSE web-
between the 6 (or 12 in the case of some of the colours can mix fluxes from
HD 204155) individual observations (Fig- the different arms, which may contribute sciops/tools/MUSE_Spectral_Library.html
ures 2 and A.1 in Ivanov et al., 2019). A to the larger scatter. 3
T he MUSE spectral library at VizieR/CDS: http://
direct comparison of the MSL and XSL
spectra for eight randomly selected stars
across the spectral type sequence is Acknowledgements
shown in select wavelength ranges in This paper is based on observations made with
Figure 2. In most cases, the agreement the ESO VLT at the La Silla Paranal Observatory
on a scale of a few hundred pixels — in (Programme ID 099.D-0623). We have made exten-
other words, within the same X-shooter sive use of SIMBAD at the Centre de Données
astronomiques Strasbourg (CDS) and of the VizieR
spectral order — is excellent. However, tool and CDS, Strasbourg, France. Enrico M. Corsini,
on a larger scale we find deviations Elena Dalla Bontà, Lorenzo Morelli, and Alessandro
between the XSL and MSL spectra. We Pizzella acknowledge financial support from Padua

18 The Messenger 178 – Quarter 4 | 2019

Enrico Sacchetti/ESO


The VLTI delay lines at Paranal lie

inside a 168-metre tunnel, forming
an essential part of a complicated
optical system that feeds interfero­
metric instruments such as GRAVITY.

The Messenger 178 – Quarter 4 | 2019 19

GRAVITY Science DOI: 10.18727/0722-6691/5166

Spatially Resolving the Quasar Broad Emission

Line Region

GRAVITY Collaboration Juan Pablo Gil 8 Anne-Lise Maire 23, 3

Stefan Gillessen 1 Leander Mehrgan 8
Roberto Abuter 8 Frédéric Gonté 8 Antoine Mérand 8
Matteo Accardo 8 Paulo Gordo 6 Florentin Millour 37
Tobias Adler 3 Damien Gratadour 2 Paul Mollière 3
António Amorim 6 Alexandra Greenbaum 40 Thibaut Moulin 5
Narsireddy Anugu 7, 29, 30 Rebekka Grellmann 4 André Müller 3
Gerardo Ávila 8 Ulrich Grözinger 3 Eric Müller 8, 3
Michi Bauböck 1 Patricia Guajardo 8 Friedrich Müller 3
Myriam Benisty 5,12 Sylvain Guieu 5 Hagai Netzer 32
Jean-Philippe Berger 5 Maryam Habibi 1 Udo Neumann 3
Joachim M. Bestenlehner 22, 3 Pierre Haguenauer 8 Mathias Nowak 2
Hervé Beust 5 Oliver Hans 1 Sylvain Oberti 8
Nicolas Blind 9 Xavier Haubois 8 Thomas Ott 1
Mickaël Bonnefoy 5 Marcus Haug 8 Laurent Pallanca 8
Henri Bonnet 8 Frank Haußmann 1 Johana Panduro 3
Pierre Bourget 8 Thomas Henning 3 Luca Pasquini 8
Jérôme Bouvier 5 Stefan Hippler 3 Thibaut Paumard 2
Wolfgang Brandner 3 Sebastian F. Hönig 27 Isabelle Percheron 8
Roland Brast 8 Matthew Horrobin 4 Karine Perraut 5
Alexander Buron 1 Armin Huber 3 Guy Perrin 2
Leonard Burtscher 14 Zoltan Hubert 5 Bradley M. Peterson 24, 25, 26
Faustine Cantalloube 3 Norbert Hubin 8 Pierre-Olivier Petrucci 5
Alessio Caratti o Garatti 16, 3 Christian A. Hummel 8 Andreas Pflüger 1
Paola Caselli 1 Gerd Jakob 8 Oliver Pfuhl 8
Frédéric Cassaing 10 Annemieke Janssen 36 Than Phan Duc 8
Frédéric Chapron 2 Alejandra Jimenez Rosales 1 Jaime E. Pineda 1
Benjamin Charnay 2 Lieselotte Jochum 8 Philipp M. Plewa 1
Élodie Choquet 37 Laurent Jocou 5 Dan Popovic 8
Yann Clénet 2 Jens Kammerer 8, 41 Jörg-Uwe Pott 3
Claude Collin 2 Martina Karl 20, 21 Almudena Prieto 39
Vincent Coudé du Foresto 2 Andreas Kaufer 8 Laurent Pueyo 11
Ric Davies 1 Stefan Kellner 1 Sebastian Rabien 1
Casey Deen 1 Sarah Kendrew 11, 3 Andrés Ramírez 8
Françoise Delplancke-Ströbele 8 Lothar Kern 8 José Ricardo Ramos 3
Roderick Dembet 8 Pierre Kervella 2 Christian Rau 1
Frédéric Derie 8 Mario Kiekebusch 8 Tom Ray 16
Willem-Jan de Wit 8 Makoto Kishimoto 31 Miguel Riquelme 8
Jason Dexter 1 Lucia Klarmann 3 Gustavo Rodríguez-Coira 2
Tim de Zeeuw 1,14 Ralf Klein 3 Ralf-Rainer Rohloff 3
Catherine Dougados 5 Rainer Köhler 3 Daniel Rouan 2
Guillaume Dubus 5 Yitping Kok 1 Gérard Rousset 2
Gilles Duvert 5 Johann Kolb 8 Joel Sanchez-Bermudez 3, 17
Monica Ebert 3 Maria Koutoulaki 16, 19, 3, 8 Marc Schartmann 1, 33, 34
Andreas Eckart 4,13 Martin Kulas 3 Silvia Scheithauer 3
Frank Eisenhauer 1 Lucas Labadie 4 Markus Schöller 8
Michael Esselborn 8 Sylvestre Lacour 2, 8 Nicolas Schuhler 8
Fabio Eupen 4 Anne-Marie Lagrange 5 Dominique Segura-Cox 1
Pierre Fédou 2 Vincent Lapeyrère 2 Jinyi Shangguan 1
Miguel C. Ferreira 6 Werner Laun 3 Thomas T. Shimizu 1
Gert Finger 8 Bernard Lazareff 5 Jason Spyromilio 8
Natascha M. Förster Schreiber 1 Jean-Baptiste Le Bouquin 5 Amiel Sternberg 1, 32
Feng Gao 1 Pierre Léna 2 Matthias Raphael Stock 21
César Enrique García Dabó 8 Rainer Lenzen 3 Odele Straub 1, 2
Rebeca Garcia Lopez 16, 3 Samuel Lévêque 8 Christian Straubmeier 4
Paulo J. V. Garcia 7 Chien-Cheng Lin 3,18 Eckhard Sturm 1
Éric Gendron 2 Magdalena Lippa 1 Marcos Suárez Valles 8
Reinhard Genzel 1,15 Dieter Lutz 1 Linda J. Tacconi 1
Ortwin Gerhard 1 Yves Magnard 5 Wing-Fai Thi 1

20 The Messenger 178 – Quarter 4 | 2019

Konrad R. W. Tristram 8 15
Department of Physics, Le Conte Hall, 40
University of Michigan Department of
Javier J. Valenzuela 8 University of California, Berkeley, USA Astronomy, Ann Arbor, USA
Roy van Boekel 3 16
Dublin Institute for Advanced Studies, 41
Research School of Astronomy &
Ewine F. van Dishoeck 14 Dublin, Ireland Astrophysics, Australian National
Pierre Vermot 2 17
Instituto de Astronomía, Universidad ­University, Canberra, Australia
Frédéric Vincent 2 Nacional Autónoma de México, Ciudad
Sebastiano von Fellenberg 1 de México, Mexico
Idel Waisberg 1 18
Institute for Astronomy, University of The angular resolution of the Very Large
Jason J. Wang 28 Hawai’i, Honolulu, USA Telescope Interferometer (VLTI) and
Imke Wank 4 19
School of Physics, University College the excellent sensitivity of GRAVITY
Johannes Weber 1 Dublin, Ireland have led to the first detection of spa-
Gerd Weigelt 13 20
Max Planck Institute for Physics, tially resolved kinematics of high veloc-
Felix Widmann 1 Munich, Germany ity atomic gas near an accreting super-
Ekkehard Wieprecht 1 21
TUM Department of Physics, Technical massive black hole, revealing rotation
Michael Wiest 4 University of Munich, Garching, on sub-parsec scales in the quasar
Erich Wiezorrek 1 Germany 3C 273 at a distance of 550 Mpc. The
Markus Wittkowski 8 22
Department of Physics and Astronomy, observations can be explained as the
Julien Woillez 8 University of Sheffield, UK result of circular orbits in a thick disc
Burkhard Wolff 8 23
STAR Institute, Liège, Belgium configuration around a 300 million solar
Pengqian Yang 3, 35 24
Department of Astronomy, The Ohio mass black hole. Within an ongoing
Senol Yazici 1, 4 State University, Columbus, USA Large Programme, this capability will be
Denis Ziegler 2 25
Center for Cosmology and AstroParticle used to study the kinematics of atomic
Gérard Zins 8 Physics, The Ohio State University, gas and its relation to hot dust in a
Columbus, USA sample of quasars and Seyfert galaxies.
Space Telescope Science Institute, We will measure a new radius-luminosity
Max Planck Institute for Extraterrestrial Baltimore, USA relation from spatially resolved data and
Physics, Garching, Germany School of Physics & Astronomy, test the current methods used to meas-
LESIA, Observatoire de Paris, Université ­University of Southampton, UK ure black hole mass in large surveys.
PSL, CNRS, Sorbonne Université, Department of Astronomy, California
­Université de Paris, Meudon, France Institute of Technology, Pasadena,
Max-Planck-Institut für Astronomie, USA Introduction
Heidelberg, Germany Steward Observatory, Department
I Physikalisches Institut, Universität zu of Astronomy, University of Arizona, Emission lines of atomic gas velocity-
Köln, Germany Tucson, USA broadened to widths of 3000–
5 30
Univ. Grenoble Alpes, CNRS, IPAG, University of Exeter, School of Physics 10 000 km s –1 are a hallmark of quasars
Grenoble, France and Astronomy, Exeter, UK and are thought to trace the gravitational
6 31
CENTRA and Universidade de Lisboa Kyoto Sangyo University, Department potential of the central supermassive
− Faculdade de Ciências, Lisboa, of Astrophysics and Atmospheric black hole. Despite decades of study
Portugal Sciences, Japan their physical origin remains unclear. The
7 32
CENTRA and Universidade do Porto – School of Physics and Astronomy, observed properties can be explained
Faculdade de Engenharia, Porto, Tel Aviv University, Israel by emission from discrete, collapsed
Portugal E xcellence Cluster Origins, Ludwig- clouds or high-density regions of a con-
ESO Maximilians-Universität München, tinuous medium. The gas may be part
Observatoire de Genève, Université de Garching, Germany of the inflow feeding the black hole or a
Genève, Versoix, Switzerland Universitäts-Sternwarte München, continuous equatorial outflow. Assuming
DOTA, ONERA, Université Paris- Munich, Germany a gravitational origin, line widths com-
Saclay, Châtillon, France Shanghai Institute of Optics and Fine bined with a measurement of the emis-
European Space Agency, Space Mechanics, Chinese Academy of sion region size provide an estimate of
­Telescope Science Institute, Baltimore, Sciences, China the black hole mass.
USA NOVA Optical Infrared Instrumentation
Unidad Mixta Internacional Franco- Group at ASTRON, Dwingeloo, the Extensive monitoring campaigns use
Chilena de Astronomía (CNRS UMI Netherlands light echoes in a technique called rever-
3386), Departamento de Astronomía, Aix Marseille Univ, CNRS, CNES, LAM, beration mapping to measure the emis-
Universidad de Chile, Las Condes, France sion size, with ongoing work expanding
Santiago, Chile Observatoire de la Côte d’Azur the sample size from tens (Kaspi et al.,
Max Planck Institute for Radio Astron- Lagrange, Boulevard de l’Observatoire, 2000; Peterson et al., 2004) to hundreds
omy, Bonn, Germany Nice, France (Du et al., 2016; Grier et al., 2017). The
14 39
Sterrewacht Leiden, Leiden University, Instituto de Astrofísica de Canarias, key result of these studies is that the
Leiden, the Netherlands La Laguna, Spain size of the emitting region increases with

The Messenger 178 – Quarter 4 | 2019 21

GRAVITY Science GRAVITY Collaboration, Spatially Resolving the Quasar Broad Emission Line Region

a Radial velocity (km s –1) b 30

– 4,000 – 2,000 0 2,000 4,000
0.8 PAJet = 222° 2.178
1.6 20 Photocentre
Pa α flux

Observed wavelength (µm)

0.6 2.176

Δ Declination (µas)
Differential phase (°)

0.4 1.2 2.174

Normalised flux
1.0 2.172
0.8 –10
0.0 2.170
Phase – 20 2.168
– 0.2
0.4 BLR model
BLR model
– 0.4 0.2 – 30
2.14 2.15 2.16 2.17 2.18 2.19 2.20 30 20 10 0 –10 – 20 – 30
Observed wavelength (µm) Δ Right ascension (µas)

c Yv d 150 3
PAJet = 222°
100 2

Radial velocity (10 3 km s –1)

Δ Declination (µas)

ev e¦
50 1

"KNTC 0 0

, %+ – 50 –1
–100 –2
1@CH@K `
–150 –3
150 100 50 0 – 50 –100 –150
1 LHM 1 !+1 Q Δ Right ascension (µas)
Figure 1. GRAVITY spatially resolves the broad emis- This is the result of net ordered rotation of the engines. The key components of AGN
sion line kinematics of 3C 273. (a) Paa line profile line-emitting gas. By comparing a kinematic model
are small on the sky, at micro- to milli-­
(black) and averaged differential phase (blue), show- of the emission region (c) to GRAVITY data, we
ing non-zero phases and a change of sign across find that a thick disc configuration viewed at low arcsecond scales, requiring long baselines
the broad emission line. (b) Photocentre positions inclination best explains the data (d). The model also at the VLTI and Keck Interferometer. AGN
measured at each line channel, showing a clear sep- provides estimates of the mean emission radius are also relatively faint sources, so far only
aration between red and blue which corresponds and central black hole mass. Adapted from GRAVITY
detected in optical interferometry with
to a velocity gradient at a position angle perpendicu- Collaboration (2018).
lar to the large-scale radio jet of 3C 273 (black line). 8–10-metre-class telescopes and instru-
mentation with excellent sensitivity. Con-
tinuum measurements with the Keck
luminosity, roughly as R ~ L1/2. That rela- campaigns to measure R via an estimate Interferometer (for example, Kishimoto et
tionship can be understood as atomic based on L). Secondary methods so far al., 2011) and the Astronomical Multi-
gas emission being produced under opti- provide all available active galactic nucleus BEam combineR (AMBER) on the VLTI
mal photoionisation conditions (constant (AGN) black hole mass measurements (Weigelt et al., 2012) provide information
received flux). This radius-luminosity rela- in large samples and out to high redshift. about hot dust surrounding the nucleus.
tion allows “secondary” methods for The broad line region (BLR) is even smaller
­estimating black hole masses using a Interferometry provides an independent (angular size < 0.1 milliarcseconds [mas])
­single optical spectrum (replacing long method for spatially resolving AGN central and is impossible to resolve in standard

22 The Messenger 178 – Quarter 4 | 2019

Figure 2. AGN radius-luminosity relationships meas-
GRAVITY BLR ured for hot dust and atomic gas. The hot dust
10 0 measurements include our new GRAVITY results
GRAVITY hot dust (purple solid circles; see GRAVITY Collaboration,
Hot dust RM/OI 2019a), as well as those from previous observations.
Gas RM For atomic gas, we have detected velocity gradients
and measured the emission region size for the
­quasar 3C 273 (GRAVITY Collaboration, 2018) with
another detection and upper limits in deep integra-
10 –1
Radius (pc)

tions for two other sources. With an ongoing large

programme, we aim to expand the sample to roughly
10 AGN spanning four orders of magnitude in
­luminosity. The results can be compared to the large
scatter found in reverberation mapping samples
­(different samples as smaller symbols) and to the
10 – 2 R ~ L0.5 relations found for both dust and atomic gas.

marily set by rotation in the black hole

gravitational potential, or by polar outflow
10 – 3 driven by radiation pressure? And is the
velocity structure well ordered or
1043 1044 1045 1046 1047
Bolometric luminosity (erg s –1)
imaging, even with the VLTI. Instead, we By adopting a kinematic model of the By modelling the line profile and differen-
can study its kinematics by measuring Paa emission region as a collection of tial phase data, we will measure the
the photocentre shift of the atomic gas orbiting gas clouds (following Pancoast et emission region size and construct a new
relative to the hot dust, as a function of al., 2014 and Rakshit et al., 2015), we radius-luminosity relationship. Our results
wavelength (or velocity) across the emis- measure physical properties of the gas can be compared with those obtained
sion line. The photocentre shift results in distribution and black hole. The data are independently from reverberation tech-
a small differential phase signal ~
< 1 degree consistent with a thick disc (opening niques and used to constrain the physical
(Rakshit et al., 2015) whose detection angle of 45+– 69 degrees) in Keplerian rota- origin of the atomic gas. We will also
requires high sensitivity and deep integra- tion around a supermassive black hole of study the connection of the atomic gas to
tions. This is now possible with GRAVITY. 1.5–4.1 × 108 M⊙. The inclination and that of the hot dust continuum which we
position angles agree with those inferred obtain using the same data (for example,
for the radio jet. The measured mean GRAVITY Collaboration, 2019a & b). The
A case study in 3C 273 emission radius of RBLR = 0.12 ± 0.03 pc angular size of both the hot dust and the
(at an angular diameter distance of atomic gas scales with optical flux, which
We observed 3C 273 with GRAVITY 548 Mpc) is a factor of about two smaller makes interferometry well suited for
using the four Unit Telescopes (UTs) over than reported in earlier RM studies (Kaspi ­studying luminous quasars like 3C 273 as
eight nights between July 2017 and et al., 2000; Peterson et al., 2004) well as nearby Seyfert galaxies. A future
May 2018, with a total on-source integra- although it is consistent with a recent one upgrade to the sensitivity of GRAVITY
tion time of 8 hours. By combining the (Zhang et al., 2019). This first result sup- could further obtain kinematics, broad
data from all epochs, we measure the ports the fundamental assumptions used emission line region size, and black hole
interferometric phase with a precision of in reverberation mapping and the sec- mass estimates for large samples out to
~ 0.1–0.2 degrees per baseline. An aver- ondary methods used to measure black a redshift z ~ 2.
age of three of the six baselines shows hole mass. For more details, see GRAVITY
the detection of an S-shaped phase sig- Collaboration (2018).
nal, corresponding to a spatially resolved Acknowledgements
velocity gradient across the otherwise This research was supported by Paris Observatory,
featureless broad Paa emission line (Fig- Outlook Grenoble Observatory, by CNRS/INSU, by the Pro-
ure 1a). From the phase data, we fit for a gramme National Cosmologie et Galaxies (PNCG)
model-independent photocentre position With an approved large programme of CNRS/INSU with INP and IN2P3, co-funded by
CEA and CNES, by the Programme National GRAM
at wavelength channels where the line we are carrying out observations of of CNRS/INSU with INP and IN2P3, co-funded by
emission is strong. We find a clear sepa- ~ 10 sources over the next two years, CNES, by the Programme National Hautes Energies
ration between blue and red channels (a spanning four orders of magnitude in (PNHE) of CNRS/INSU with INP and IN2P3, co-funded
velocity gradient, Figure 1b), with an ori- AGN luminosity. The data will provide by CEA and CNES, and by the Programme National
de Physique Stellaire (PNPS) of CNRS/INSU, co-
entation perpendicular to the large-scale information on the dominant kinematics funded by CEA and CNES. It has also received fund-
radio jet. This demonstrates net rotation and the degree of ordered motion in ing from the following programmes: European
of the line emission region. The photo- atomic gas in the broad emission line Union’s Horizon 2020 research and innovation pro-
centre positions are measured with a typ- region, helping us to address the follow- gramme (OPTICON Grant Agreement 730890), from
the European Research Council (ERC) under the
ical precision of 5 µas per channel. ing questions: are the line widths pri­

The Messenger 178 – Quarter 4 | 2019 23


European Union’s Horizon 2020 research and inno- FIS/00099/2013, SFRH/BSAB/142940/2018 [P. G.] GRAVITY Collaboration 2018, Nature, 563, 657
vation programme (Grant Agreement No. 743029), and PD/BD/113481/2015; M. F. in the framework of GRAVITY Collaboration 2019a, submitted to A&A,
from the Irish Research Council (IRC Grant: the Doctoral Programme IDPASC Portugal), by NSF arXiv:1910.00593
GOIPG/2016/769) and SFI Grant 13/ERC/12907, grant AST 1909711, by the Heising-Simons Founda- GRAVITY Collaboration 2019b, submitted to A&A
from the Humboldt Foundation Fellowship and the tion 51 Pegasi b postdoctoral fellowship, from the Bentz, M. C. et al. 2013, ApJ, 767, 149
ESO Fellowship programes, from the European Direction Scientifique Générale of Onera and by a Du, P. et al. 2018, ApJ, 856, 6
Research Council under the European Union’s Hori- Grant from Science Foundation Ireland under Grant Grier, C. J. et al. 2017, ApJ, 851, 21
zon 2020 research and innovation programme (Grant number 18/SIRG/5597. Pancoast, A. et al. 2008, MNRAS, 445, 3073
Agreement Nos. 2016–ADG–74302 [EASY], 2015- Kishimoto, M. et al. 2011, A&A, 527, 121
StG-677117 [SFH], 694513, and 742095 [SPIDI]), and Weigelt, G. et al. 2012, A&A Letters, 451, 9
was supported in part by the German Federal Minis- References Zhang, Z.-X. et al. 2019, ApJ, 876, 49
try of Education and Research (BMBF) under the
grants Verbundforschung #05A08PK1, #05A11PK2, Peterson, B. M. et al. 2004, ApJ, 613, 682
#05A14PKA and #05A17PKA, by Fundação para a Kaspi, S. et al. 2000, ApJ, 533, 631
Ciência e a Tecnologia, Portugal (Grants UID/ Rakshit, S. et al. 2015, MNRAS, 447, 2420

DOI: 10.18727/0722-6691/5167

An Image of the Dust Sublimation Region in the

Nucleus of NGC 1068

GRAVITY Collaboration (see page 20) Since the first seminal paper addressing Under superb conditions, with seeing
its physical properties (Krolik & ­Begelman, ~ 0.5 arcseconds and a coherence time
1988), and following numerous observa- of up to 13 ms, it was possible to
The superb resolution of the Very Large tions at many different wavelengths, ­fringe-track on the nucleus of NGC 1068
Telescope Interferometer (VLTI) and the “torus” concept has evolved and despite its large size and moderate
the unrivalled sensitivity of GRAVITY been modified considerably. At the same brightness. The data obtained were of
have allowed us to reconstruct the first time, increases in computational power excellent quality, with typically < 1% visi-
detailed image of the dust sublimation have facilitated detailed modelling of bility and closure-phase accuracy. The
region in an active galaxy. In the nearby clumpy torus structures. Such models wealth of information provided by the six
archetypal Seyfert 2 galaxy NGC 1068, are consistent with the near- to mid- VLTI baselines has enabled us to recon-
the 2 µm continuum emission traces infrared spectral energy distribution as struct a K-band image based on the
a highly inclined thin ring-like structure well as dust reverberation measurements. obtained closure phases and visibilities
with a radius of 0.24 pc. The observed Observations of almost two dozen with 3-milliarcsecond (mas) resolution.
morphology challenges the picture of a ­galaxies using the MID-infrared Interfero-
geometrically and optically thick torus. metric instrument (MIDI) on the VLTI have We used the publicly available Multi-­
resolved the 1–3 pc scales where warm aperture image Reconstruction Algorithm
dust is responsible for the mid-infrared (MiRA; Thiébaut, 2008) to generate the
Introduction continuum (Burtscher et al., 2013 and ref- image shown in Figure 1, which contains
erences therein). However, measuring a total flux of 155 mJy. The structures
NGC 1068 is one of the best studied the size of the small (< 1 pc) region con- present are robust, having been repro-
nearby active galactic nuclei (AGN), in taining hot dust that emits at near-infra- duced consistently over a wide variety of
which accretion onto a central super- red wavelengths has been possible in parameter settings, and with a signal
massive black hole contributes a signifi- very few galaxies. Also, until GRAVITY level much higher than that expected for
cant fraction of the galaxy’s total luminos- observed NGC 1068, there were no data spurious sources. Full details are in
ity. The observation of broad polarised showing spatial structure in this dust sub- GRAVITY Collaboration (2019).
emission lines by Antonucci & Miller limation region.
(1985) in the nucleus of this Seyfert gal-
axy was central to the development of A new view of NGC 1068
the unified model that explains the differ- Observations and
ences between Seyfert 1 and Seyfert 2 Image Reconstruction The image in Figure 1 is dominated
objects as being due to the presence by knots of continuum arranged in a ring
of a nuclear equatorial structure that both Data on NGC 1068 were obtained in around a central hole, with the south-
obscures and scatters the central emis- November and December 2018 using western side about a factor of two brighter
sion depending on the line of sight. GRAVITY and the four 8-metre UTs. than the north-eastern side. Fitting an

24 The Messenger 178 – Quarter 4 | 2019

(X-ray) NASA/CXC/MIT/C. Canizares, D. Evans et al.,

Maser disc
(optical) NASA/STScI, (radio) NSF/NRAO/VLA

− 300, 0, 300 km s –1
10 lar
t flo

Declination offset (mas)

5 ne

0 (mJy beam −1)


1 pc
−15 0.
1600 pc 15 10 5 0 −5 −10 −15
Right ascension offset (mas)

ellipse to these knots yields a position the population of masing molecules. In Figure 1. Left: Three-colour image of NGC 1068. The
optical emission is shown in green, the X-ray in red
angle of 50 degrees west of north, an an additional test, we have compared the
and the radio jet in blue. Right: Reconstructed image
inclination of 70 degrees and a radius of spectral energy distributions predicted of the 2 µm continuum (blue colour scale) in the cen-
about 0.24 pc. The size matches remark- by models with the photometry from MIDI tral 2.1 pc of NGC 1068, showing the reconstructed
ably well the expected dust sublimation and GRAVITY. For reasonable parameter beam size in the lower left. The white dashed ellipse,
fitted to the brightest knots, traces a ring that
radius for large graphite grains in the ranges, the models tend to over-predict
matches the expected range of dust sublimation
radiation field of an AGN with an intrinsic the mid-infrared continuum and have radii (orange dotted ellipses). The filled black circle in
bolometric luminosity of ~ 4 × 1045 erg a near-infrared slope that is too shallow. the centre of the ring, denoting the location of the
s –1 as expected for NGC 1068. And if one AGN, has been matched to the kinematic centre
derived from the maser kinematics, and hence fixes
aligns the central hole in the near-infrared As an alternative, we considered whether
the relative position of the maser distribution. The
continuum to the location of the central the mid- and near-infrared continua have radio continuum has been positioned using the
black hole inferred from the maser kine- a common origin at all. Cool (~ 700 K) masers as a coordinate reference. The green dashed
matics (Gallimore &­Impellizzeri, 2019), dust behind a screen of extinction pro- lines outline the bipolar ionised outflow also seen in
polarisation data. The grey dashed ellipse indicates
then the positions of the lower-velocity vides an unexpectedly good fit to the
the size of the 10-metre continuum in the MIDI data.
maser spots match up remarkably well spectral energy distribution, including the
with the south-western side of the ring. silicate dip. But the modest AK ~ 0.9
This suggests that the masers and the magnitude extinction is far less than the requires that most of the mid-infrared
hot dust trace a common disc, and hence lower limit of AK ~ 6 magnitudes required continuum originates in a different struc-
that the brighter south-western side of by the non-detection of broad Bra ture on larger scales. Disc-plus-wind
the ring is the near side. This geometry is at 4 µm (Lutz et al., 2000) and the high- models such as those described by
consistent with that implied by the jet and column density implied by the HCN1–0 Hönig (2019) would imply that the other
the ionisation cone, which are oriented emission at 3 mm (García-Burillo et al., structure is in fact the outflow driven by
toward us on the northern side. 2016; Imanishi et al., 2018). the AGN.

The near-infrared continuum is very diffi- Instead, our preferred interpretation is in

cult to reconcile with geometrically terms of a hot dust disc close to the sub- Conclusion
thick clumpy torus models, which can limation temperature. Dust at 1500 K
only reproduce a ring-like structure in behind a screen with AK ~ 5.5 magnitude K-band observations with GRAVITY at a
systems that are relatively face-on, and extinction is able to reproduce the slope spatial resolution of 3 mas have resolved
struggle to make the near side of the of the near-infrared continuum. And a a ring-like structure on sub-parsec scales
ring brighter. Similarly, the presence of a modest scale height of h/r < 0.14, as indi- in the centre of NGC 1068. These obser-
thin maser disc is inconsistent with a ver- cated by the data, is sufficient to couple vations do not support ideas of a
tically extended structure, since this the AGN luminosity to the dust disc geometrically and optically thick clumpy
would impede the escape of far-infrared because of the misalignment between it torus and instead trace a dusty disc
photons that would otherwise thermalise and the accretion disc. This scenario around the AGN. The size matches that

The Messenger 178 – Quarter 4 | 2019 25


expected for the dust sublimation region,

and the apparent orientation is similar

to that of the maser disc, arguing for

a common origin. The structure and pho-
tometry are consistent with dust at

~ 1500 K behind AK ~ 5.5 magnitudes of

foreground extinction. This matches what

is expected from the upper limit to the
broad Bra line, and could originate in the
dense and turbulent gas distribution
observed on scales of 1–10 pc. In such a
scenario, much of the mid-infrared con- s
tinuum would originate in a separate r clo
structure, likely associated with the AGN- M Thic k molecular gas disc
driven outflow.

Acknowledgements isc
i nd
t th
See page 23. Ho
u st )
MID -IR d 5 (p
References R
0. 2
Antonucci, R. R. J. & Miller, J. S. 1985, ApJ, 297, 621
Burtscher, L. et al. 2013, A&A, 558, 149 0
García-Burillo, S. et al. 2016, ApJ, 823, L12
Gallimore, J. & Impellizzeri, V. 2019, submitted to
ApJ Figure 2. Sketch of the observed central structures. from the disc periphery. ALMA observations of HCN
GRAVITY Collaboration 2019, accepted by A&A The K-band emission traces the inner rim of a thin and HCO+ show a turbulent structure, which rotates
Hönig, S. F. 2019, accepted by ApJ disc of hot gas and dust, at or close to the dust sub- in the opposite direction to the maser disc (Imanishi
Imanishi, M. et al. 2018, ApJ, 853, L25 limation radius of 0.24 pc. The inner water masers et al., 2018). The turbulence found in the molecular
Krolik, J. H. & Begelman, M. C. 1988, ApJ, 329, 702 are cospatial with the hot K-band dust. The masers gas structure argues for a thick disc, which contains
Lutz, D. et al. 2000, ApJ, 530, 733 stretch out to 1 pc (Gallimore et al., 2001). Mid-infra- enough gas mass to reach column densities
Thiébaut, E. 2008, Proc. SPIE, 7013, 70131I red observations show warm dust on roughly the that screen the central region from the observer by
same scales as the outer masers, likely originating AK ~ 5.5 magnitudes.

DOI: 10.18727/0722-6691/5168

GRAVITY and the Galactic Centre

GRAVITY Collaboration (see page 20) vitational redshift, the most precise mass- It is embedded in hot gas and sur-
distance measurement, the test of the rounded by a cluster of high velocity
equivalence principle, and the detection stars. They buzz around the black
On a clear night, our home galaxy, the of orbital motion near the black hole. hole on trajectories which are, like the
Milky Way, is visible as a starry ribbon behaviour of the hot gas, governed by
across the sky. Its core is located in the gravitational field of the black hole.
the constellation of Sagittarius, approx- The heart of the Milky Way
imately where the bright glow is inter- With GRAVITY we are unravelling what is
rupted by the darkest dust filaments. At the heart of the Milky Way, 26 000 light- happening in the centre of our Galaxy
There, hidden, lies a massive black hole. years from Earth, is Sagittarius A* (Sgr A*, with unprecedented angular resolution.
To peer through the obscuring clouds pronounced “Sag-A-star”), the closest The instrument operates at infrared
and see the stars and gas near the massive black hole to us and, with a wavelengths around 2 microns. GRAVITY
black hole we use GRAVITY. The main lensed angular diameter of 53 microarc­ combines the light beams of the four
GRAVITY results are the detection of gra- seconds (µas), the largest one on the sky. ­individual 8.2-metre Unit Telescopes at

26 The Messenger 178 – Quarter 4 | 2019

Y. Beletsky/ESO

ESO/MPE/S. Gillessen et al.

ESO’s Very Large Telescope (VLT) in fields of Earth, the Sun, and white dwarfs. Figure 1. Left: The sky above the VLT at Paranal. The
laser of Unit Telescope 4 (Yepun) points at the
Chile to form the VLT Interferometer With GRAVITY and SINFONI we were
Galactic centre. Right: Infrared image of the Galactic
(VLTI). Together they achieve a spatial able to test the strong gravitational field centre. For the interferometric GRAVITY observa-
resolution equivalent to that of a tele- of a massive black hole. During its recent tions the star IRS 16C was used as a reference star
scope of approximately 130 metres in closest approach to Sgr A* the spectral and the actual target was the star S2. The position of
the centre, which harbours the (invisible) 4 million
diameter (GRAVITY Collaboration, 2017). absorption lines in the light of S2 were
solar mass black hole known as Sgr A* is marked by
GRAVITY has also been equipped with significantly shifted towards redder the orange cross.
a system to track interference fringes ­wavelengths, in excellent agreement with
and it uses adaptive optics to correct for ­Einstein’s general theory of relativity The star S2 experiences very strong
atmospheric turbulence in order to (GRAVITY Collaboration 2018a). changes in gravitational potential in the
resolve small and faint structures in the course of its eccentric orbit around
sky. Sgr A*. This makes it a unique probe and
Mass and Distance of the Galactic allows us to test the LPI. The spectrum
Black Hole of S2 has absorption lines of helium and
Measurement of gravitational redshift in hydrogen, which are formed by atomic
the Galactic centre Our measurements of the position and processes and are thus non-gravitational.
radial velocity of S2 allow us to calculate We can observe how they change in
We have traced a partial astrometric both the mass of the black hole and the wavelength as the star moves on its tra-
and a full 16-year radial velocity orbit of distance to the Galactic centre with jectory towards us, around the black
the star S2 with GRAVITY on the VLTI unprecedented precision and accuracy. hole, and away from us again. During the
and the Spectrograph for INtegral Field By combining the precise astrometry pericentre passage both the hydrogen
Observations in the Near Infrared from GRAVITY with the spectral meas- and helium lines are redshifted. We did
­(SINFONI) on the VLT. During its recent urements of SINFONI, we can determine not detect a different shift of the two
­closest approach to the black hole, the the distance to the Galactic centre to absorption lines. This puts a limit on the
pericentre passage in May 2018, we be 26 673 light-years and the black hole violation of the LPI to below 5%. While
­collected both astrometric and spectro- mass to be 4.1 million solar masses current tests on Earth have a much
scopic data. These data allowed the (GRAVITY Collaboration, 2019b). higher accuracy, our experiment in the
detection of the combined gravitational Galactic centre laboratory tests gravita-
redshift and transverse Doppler effect on tional field changes a million times larger
S2 for the first time. Gravitational redshift Local position invariance (GRAVITY Collaboration, 2019a).
is one of the three classical tests of
­Einstein’s general theory of relativity. One of the cornerstones of general rela-
­Einstein was the first to accurately predict tivity is Einstein’s equivalence principle. Flares
a gravitational time dilation, i.e., that a It consists of three parts: the weak equiv-
clock near a gravitational mass ticks alence principle, the local Lorentz invari- The Galactic centre black hole is, given
slower than a distant reference clock. As ance and the local position invariance its huge mass, surprisingly faint. That is,
a result of this effect an observer sees (LPI). We use the orbit of S2 to test the the hot gas that swirls around it has a
a photon emitted near a massive object LPI, which states that the results of a comparatively low luminosity. Most of the
at a longer, redder wavelength. This pre- non-gravitational experiment are inde- radiation is emitted at radio and infrared
diction has so far only been tested in pendent of the position in space-time. wavelengths and is quasi-steady — it
weak gravity regimes like the gravitational flickers only a little. In the near-infrared

The Messenger 178 – Quarter 4 | 2019 27

GRAVITY Science GRAVITY Collaboration, GRAVITY and the Galactic Centre


0.175 3000


vLSR (km s –1)


0.125 0

– 1000
Δ Declination (ೀ)

– 2000
2000 2005 2010 2015 2020
0.075 Time (year)


Residuals vLSR (km s –1)




0.05 0.025 0 – 0.025 – 0.05 – 0.075 50

Δ Right ascension (ೀ)

Figure 2. Orbit of S2. Astrometric data from GRAVITY 0

(blue), NACO and SHARP (red). The black ellipse is
the best-fit orbit and the black circle shows the posi-
2016.5 2017.0 2017.5 2018.0 2018.5 2019.0
tion of Sgr A*. Flare positions are marked by grey
crosses. Top right: S2 radial velocity (along our line Time (year)
of sight) measured over more than one orbit. Bottom
right: The combined gravitational redshift and relativ-
istic transverse Doppler effect manifest in an excess
in the radial velocity of 200 km s –1.
Background image: ESO/GRAVITY Consortium/L. Calçada

where GRAVITY and SINFONI operate,

the long-term light curves are well
described by a log-normal noise, indicat-
ing that there are statistical fluctuations 100
in the way the hot gas is accreted by
the black hole. On average, about once
per day for 1–2 hours this slightly variable
y-offset (µas)

emission becomes a bright flare, and

at times it contains so much energy that
it even emits X-rays. The true nature of 0
these flares seen at infrared and X-ray 7 Rg
wavelengths is not yet known and may
be explained as a hot spot in the gas or
an ejected blob of gas (as in a jet).
Figure 3. Projected orbit
Observations during the summer of 2018 –100
of the flare recorded on
with GRAVITY revealed that the emission 22 July 2018 over its
near the black hole during an infrared 30-minute duration (col-
flare moves in a loop around an unseen our ranging from brown
to dark blue indicates
centre (GRAVITY Collaboration, 2018b). the time). The back-
These loops are typically a few times 100 0 –100 ground shows a flare
larger than the event horizon of the black x-offset (µas) “hot spot” simulation.

28 The Messenger 178 – Quarter 4 | 2019


hole and are consistent with a small What’s next? properties, for example, the sense of
region of heated electrons (a “hot spot”), rotation of the hot gas.
moving in an orbit around the black hole. Continuing observations of S2 are
The GRAVITY observations also revealed expected to reveal a second relativistic
changes in the polarisation angle over effect on the star’s orbit, namely the References
the course of the flare. In particular, as Schwarzschild precession. General rela- GRAVITY Collaboration 2017, A&A, 602, 23
the centroid of the emission region com- tivity predicts that the orbit of S2 is not GRAVITY Collaboration 2018a, A&A, 615, 15G
pletes one orbit around the black hole, a closed Keplerian ellipse but an open GRAVITY Collaboration 2018b, A&A, 618, 15
the polarisation angle also makes a single rosette-like trajectory, where the peri- GRAVITY Collaboration 2019a, Phys. Rev. Lett., 122,
loop. These polarisation measurements apse, i.e., the closest point to the black GRAVITY Collaboration 2019b, A&A, 625, 10
indicate the presence of a strong mag- hole, shifts by a small angle per revolution
netic field in the immediate vicinity of the which rotates the ellipse over time.
black hole and might indicate a magnetic ­Moreover, studying multiple flares as an
origin of the flare. ensemble will shed light on accretion

DOI: 10.18727/0722-6691/5169

Spatially Resolved Accretion-Ejection in

Compact Binaries with GRAVITY

GRAVITY Collaboration (see page 20) rable to the binary orbit, had remained GRAVITY observations of two such
unresolved for a long time because the objects: the hypercritical accretor and
required sub-milliarcsecond spatial reso- exotic microquasar SS433, and the
The GRAVITY instrument at the Very lution is significantly beyond the diffraction wind-accreting high-mass XRB BP Cru.
Large Telescope Interferometer has led limit of even extremely large telescopes.
to the first spatially resolved observa- Resolving these structures is, in fact,
tions of X-ray binaries at scales compa- challenging even for optical interferometry, Resolving super-Eddington outflows in
rable to the binary orbit, providing since these sizes are below the canonical SS433
unprecedented spatial information on spatial resolution of an optical interferom-
their accretion-ejection mechanisms. eter such as the Very Large Telescope SS433 is unique in the Galaxy as the
In particular, observations of the hyper- Interferometer (VLTI), which is around 3 only known steady hypercritical accretor;
critical accretor SS433 have revealed a mas for a baseline of 100 metres. There- the donor star provides the compact
variety of spatial structures at the heart fore, in order to get to such scales, exqui- object (the nature of which remains enig-
of this exotic microquasar, including site precision in the interferometric observ- matic, but is likely to be a black hole) with
bipolar outflows, super-­Keplerian equa- ables is required, which is best achieved matter at a rate hundreds of times above
torial outflows and extended baryonic with spectrally resolved measurements Eddington (see, for example, Fabrika,
jets photoionised by collimated ultravio- using strong emission lines. This tech- 2004 for a review of SS433). The resulting
let radiation. nique is called spectral differential inter- geometrically and optically thick super-
ferometry and it can be used to acquire critical accretion disc thermally down-
robust velocity-resolved microarcsecond grades the X-ray radiation produced
X-ray binaries (XRBs) are composed of a (µas) spatial information. close to the compact object (and typically
compact object (neutron star or black seen in ordinary X-ray binaries) to ultra­
hole) accreting matter from its donor star. GRAVITY has led to a breakthrough in violet (UV) and optical wavelengths, turn-
The accretion process leads to a variety the ability to fringe-track on faint objects, ing the compact object into an accretion-
of inflow-outflow structures such as discs, allowing interferometric quantities to powered quasi-star that outshines its
streams, winds and jets. While large- be measured in the near-infrared (NIR) donor star at all wavelengths. In addition,
scale jets are often resolved with very at high spectral resolution (R ~ 4000) the enormous radiation pressure leads
long baseline interferometry (VLBI) at with unprecedented precision. When to powerful outflows producing strong
radio wavelengths, capable of achieving applied to X-ray binaries, this has led to emission lines, seen not only from the
approximately milliarcsecond (mas) spatial the first spatially resolved observations ~ 2000 km s –1 accretion disc winds (the
resolution, the inner parts of the accre- of accretion-ejection structures at NIR so-called “stationary” lines) but also
tion-ejection structures, at scales compa- wavelengths. Here, we review pioneering from the ~ 80 000 km s –1 (0.26c) highly

The Messenger 178 – Quarter 4 | 2019 29

GRAVITY Science GRAVITY Collaboration, Accretion-Ejection in Compact Binaries with GRAVITY

a) Brδjet1 Brγ Brγjet1 Brγjet2 Je
2.25 t in
He I He I jet1 He I jet2 N te n
2.00 sit

E -1

1.2 s
UT1–UT3 - 1/
Normalised visibility



Normalised phases (degrees)

10 0.2
UT1–UT3 6c

0 1 mas

> 15 mas
2.05 2.10 2.15 2.20 2.25 2.30
λ (μm)

Figure 1. Left: Spectrum (a), differential visibility ary” Brg line alternates between a bipolar- nation with spectroscopic observations,
amplitudes (b), and visibility phases (c) across the
outflow dominated mode (aligned with have shown that SS433 is UV-dominated
“stationary” and baryonic jet emission lines of
SS433 in the 2016 GRAVITY observation. The best- the jets) to an equatorial-outflow domi- even in the jet funnel. This is important
fit exponential model for the jet emission (illustrated nated mode (perpendicular to the jets) in the context of the acceleration mecha-
in the schematic on the right) is shown in black. (Figure 2). Although the presence of nism of the ~ 0.26c jets by line-locking
equatorial outflows in SS433 had been (Milgrom, 1979), which requires intense
collimated baryonic jets, with the latter’s well established from radio observations collimated radiation, as well as in the
precession creating its idiosyncratic (for example, Blundell et al., 2001), this possible relation between SS433 and
­moving lines across the X-ray and optical is the first time that velocity and size ultraluminous X-ray sources (ULXs).
spectra (Margon et al., 1979). could be combined to show that the out- Future observations of several hours in a
flows are super-Keplerian. The outflows single night could directly detect the
These unique properties — a very bright support the conclusion that both the ­relativistic motion (8 mas d –1) of the bary-
accretion disc at optical wavelengths compact object and the donor star in onic jets, providing a further probe of
and strong, broad and variable emission SS433 overfill their Roche lobes signifi- their heating mechanism and an accu-
lines — make SS433 the ideal XRB for cantly, losing mass through their outer rate, self-consistent distance to SS433.
NIR interferometry, also providing the Lagrangian points, and that the transfer
only opportunity to spatially study a of specific angular momentum between
supercritical accretion disc and its out- the binary and the disc-like outflows is Spatially resolved wind accretion in
flows. GRAVITY observations of SS433 very significant for the binary evolution. BP Cru
carried out in 2016 and 2017 (GRAVITY Future observations of several hours in a
Collaboration et al., 2017b; Waisberg et single night could harvest the full power BP Cru (GX 301-2) is composed of an
al., 2019a,b) have revealed a marginally of aperture synthesis and provide velocity- X-ray pulsar accreting from the wind of its
resolved NIR continuum consisting of (i) resolved, model-independent images hypergiant B1Ia+ companion (Kaper et
the central unresolved binary (with a size of the complex outflow structure beyond al., 2006). The latter has unusually power-
< 0.5 astronomical units [au]), and (ii) simple geometric models. ful winds (~ 10 – 5 M⊙ yr –1) for a donor star
extended emission of size ~ 40 au in the in an XRB, which lead to strong emission
form of a wind and/or disc (contributing The GRAVITY observations have also lines of HeI and Brg from its extended
~ 20% of the K-band flux). Much more spatially resolved the optical jets of wind in its K-band spectrum. In addition,
information, however, is gathered from SS433 for the first time, revealing expo- its unusually high eccentricity (e = 0.46)
the spectrally resolved differential visibility nential profiles that extend to over 20 au makes it an ideal target for probing the
amplitudes and phases across the many (i.e., several tens of times the binary size) influence of the gravitational and radiation
emission lines (Figure 1). and which peak surprisingly close to the fields of the pulsar on the surrounding
central binary (Figure 1). These observa- ­circumstellar environment (for example,
For instance, the observations have tions suggest that optical jets are heated Blondin, 1994).
shown that the double-peaked “station- by collimated UV radiation and, in combi-

30 The Messenger 178 – Quarter 4 | 2019


0.8 0.8
17 July 2016 9 July 2017
Bipolar outflow Equatorial Outflow
0.6 0.6

0.4 0.4
Brγ stationary

Declination (mas)
Declination (mas)

–1000 0 1000 Brγ stationary

0.2 v (km s –1) 0.2
–1000 0 1000
v (km s –1)
Approaching jet
0.0 0.0

Approaching jet
– 0.2 – 0.2 Receding jet

– 0.4 0.2 mas ≈ 1 au – 0.4 0.2 mas ≈ 1 au

Receding jet

– 0.6 – 0.6
1.00 0.75 0.50 0.25 0.00 – 0.25 – 0.50 – 0.75 –1.00 1.00 0.75 0.50 0.25 0.00 – 0.25 – 0.50 – 0.75 –1.00
Right ascension (mas) Right ascension (mas)

Figure 2. Velocity-resolved emission centroids of the stellar wind facing the compact Acknowledgements
across the double-peaked Brg “stationary” line for
object. In addition, asymmetries revealed
observations in 2016 (left) and 2017 (right). The See page 23.
­e mission centroid of the spatially resolved baryonic by the differential visibility phases across
jets is also shown. The black circles correspond to the emission lines may point to an addi-
the estimated binary orbit size. tional component, possibly a stream of References
enhanced density which has been pos-
Blondin, J. M. 1994, ApJ, 435, 756
The spectral differential visibilities meas- ited to exist in the system from the analy- Blundell, K. et al. 2001, ApJ, 562, L79
ured by GRAVITY (GRAVITY Collabora- sis of X-ray light curves (Leahy & Kotska, Fabrika, S. 2004, Space Science Reviews, 12, 1
tion et al., 2017a) reveal an extended wind 2008). Further observations at different GRAVITY Collaboration et al. 2017a, ApJ, 844, 177
GRAVITY Collaboration et al. 2017b, A&A, 602, L11
with a size several times the stellar radius, orbital phases could take advantage of
Kaper, L. et al. 2006, A&A, 457, 595
which is also significantly distorted — the significant eccentricity in order to Leahy, D. & Kostka, M. 2008, MNRAS, 384, 747
being more extended on the side that is ­disentangle intrinsic variability of the wind Margon, B. et al. 1979, ApJ, 233, L63
shielded from the pulsar — and which from the distortion caused by the pulsar Milgrom, M. 1979, A&A, 78, L9
Waisberg, I. et al. 2019a, A&A, 623, A47
could be caused by the X-ray ionisation accretion.
Waisberg, I. et al. 2019b, A&A, 624, A127

DOI: 10.18727/0722-6691/5170

Images at the Highest Angular Resolution with GRAVITY:

The Case of h Carinae

GRAVITY Collaboration (see page 20) massive binary η Car with GRAVITY from the detection of planets to mapping
across two spectral lines: He I and Brg. the cores of active galactic nuclei (AGN).

The main goal of an interferometer is to Interferometers reach a level of detail pro-

probe the physics of astronomical Interferometric imaging portional to the separation between each
objects at the highest possible angular pair of telescopes in the array, known
resolution. The most intuitive way of With a resolving power that is a factor of as baselines. Baselines record informa-
doing this is by reconstructing images tens of times better than stand-alone tion, at a given orientation, of the bright-
from the interferometric data. GRAVITY ­telescopes, infrared interferometry offers ness distribution of the object on the sky.
at the Very Large Telescope Interferom- the possibility to produce milliarsecond Interferometric observables, called visibil-
eter (VLTI) has proven to be a fantastic (mas) resolution images. Therefore, inter- ities, are a series of Fourier (spatial) fre-
instrument in this endeavour. In this ferometric imaging is a key means to quencies. These frequencies correspond
article, we describe the reconstruction acquire information addressing a broad to different levels of detail in the image.
of the wind-wind collision cavity of the range of astronomical problems, ranging The highest frequencies trace the finest

The Messenger 178 – Quarter 4 | 2019 31

GRAVITY Science GRAVITY Collaboration, The Case of h Carinae

Figure 1. Upper-right η Car Brγ images (February 2016, Φ = 1.28)

panel: Composite image
of the “Homunculus – 693.0 km s –1 – 657.0 km s –1 – 620.0 km s –1 – 584.0 km s –1
Nebula” taken with the
Wield Field and Plane-
tary Camera 2 (WFPC2)
on board the Hubble
Space Telescope.
Small panels: Brg recon-
structed images from – 547.0 km s –1 – 510.0 km s –1 – 474.0 km s –1 – 437.0 km s –1
the Feb. 2016 data.
With the Homunculus
Nebula having a pro-
jected size of 17 arcsec,
the GRAVITY images
represent an 850 times
zoom into the core of – 401.0 km s –1 – 364.0 km s –1 – 327.0 km s –1 – 291.0 km s –1
h Car. The Doppler
velocity of each frame
is labeled in the images.
The coloured squares
in the images represent
the different spectral
channels across the line
– 254.0 km s –1 – 218.0 km s –1 –181.0 km s –1 –144.0 km s –1 –108.0 km s –1 –71.0 km s –1 – 35.0 km s –1
that were reconstructed.

2.0 km s –1 39.0 km s –1 75.0 km s –1 112.0 km s –1 148.0 km s –1 185.0 km s –1 222.0 km s –1

258.0 km s –1 295.0 km s –1 331.0 km s –1 368.0 km s –1 405.0 km s –1 441.0 km s –1 478.0 km s –1

Declination (mas)

514.0 km s –1 551.0 km s –1 3
Norm. flux

Wavelength (m)
–10 1
2.162 2.164 2.166 2.168 2.170
10 0 –10
Right ascension (mas)

textures (like granular surfaces, or point- These algorithms minimise (i) the differ- The massive binary at the core of h Car
like objects) while the lowest ones trace ence between the data and the visibilities
extended textures (like edges and con- obtained from the model image (i.e., Located at the core of the “Homunculus
tours). An image is, therefore, composed the likelihood term), and (ii) the value of Nebula” (see Figure 1) at a distance of
of an infinite number of frequencies. one or several priors (i.e., the regularis- 2.3 kiloparsecs, h Car is a very massive
However, interferometers only sample a ers), which are defined based on the and intriguing object. Indirect observations
few of them. knowledge of the source (see Sanchez-­ suggest that a binary with a period of
Bermudez et al., 2018). Reconstruction 5.54 years resides in its core. The ­primary,
Thus, recovering an image from interfero- packages available to the community hA, is supposedly a star with a mass of
metric data is an “ill-posed” problem with optimise through gradient-descent (for around 100 M⊙, while the secondary, hB,
more unknowns (pixels in the image) than example, MiRA: Thiébaut, 2008; BSMEM: appears to be a hotter star, perhaps a
constraints (data). Reconstructing images Buscher, 1994) or Monte-Carlo Markov- giant O-star, with a mass of around 30 M⊙,
requires the use of iterative regularised Chain methods (for example, SQUEEZE: but around 100 times fainter than the pri-
least-squares minimisation algorithms. Baron & Kloppenborg, 2010). mary. Different observations suggest that

32 The Messenger 178 – Quarter 4 | 2019

AMBER – 2014 GRAVITY – 2016 GRAVITY – 2018 Figure 2. Comparison between the wind-wind colli-
sion zone’s morphologies at different orbital phases
15 of hB taken with AMBER (2014) and GRAVITY (2016,
Milliarcseconds (mas)

10 2018) at a Doppler (blue-shifted) velocity of

– 280 km s – 1. The projected trajectory of the second-
5 ary and its position are marked by the green ellipse
and dot, respectively, in each of the panels. Notice
0 how the structure of the cavity changes considerably
–5 depending on the secondary’s orbital phase. For
example, the south-eastern clump observed in 2016
–10 disappears in the 2018 reconstruction.
10 0 –10 10 0 –10 10 0 –10
Milliarcseconds (mas) Milliarcseconds (mas) Milliarcseconds (mas)

hA exhibits a very dense and slow wind was therefore selected as a target for ionised He I is formed from (i) a portion of
that shocks with a much faster and the Guaranteed Time Observations (GTO) the primary wind, which is ­photoionised
lighter wind from the secondary. GRAVITY programme (GRAVITY Collabo- by the strong ultraviolet radiation of the
ration, 2017), with the objective of carry- hB wind, and (ii) by the shocked material
The existence of hB produces several ing out a long-term (monitoring) analysis in the cavity walls. To properly quantify
changes in the morphology of hA’s wind. of the wind-wind collision cavity through this scenario new spectro-interferometric
In particular, it photoionises part of the interferometric imaging. images are required in combination with
primary wind, changing the strength of dedicated modelling. Two additional
lines such as Ha, He I, Fe II, or Ne II The first reconstructed images, pre- imaging epochs in 2018 and 2019 have
­(Mehner et al., 2010, 2012; Madura et al., sented in GRAVITY Collaboration (2018), been obtained with GRAVITY. From the
2012). 2D radiative transfer models and included data taken during the commis- preliminary analysis of data taken in 2018,
3D hydrodynamical simulations of the sioning phase (in 2016) of GRAVITY we can confirm that the morphology of
wind-wind collision scenario suggest that and through regular P100 programmes the wind-wind collision zone changes
the high-velocity secondary wind pene- (in 2017). h Car was observed with depending on the orbital phase of the
trates the slow and dense primary wind the Auxiliary Telescopes (ATs) using the secondary (Figure 2). As demonstrated in
creating a low-density cavity in it, with high-spectral-resolution mode of the case of h Car, GRAVITY spectro-in-
thin and dense walls where the two winds ­GRAVITY. This setup allowed us to terferometric imaging provides unique
interact (Madura et al., 2013; Clementel resolve several spectral lines across the information that can help to characterise
et al., 2015a,b). target’s spectrum and thereby to monitor the physics associated with the morphol-
the morphologies of the core at different ogy of complex systems at the highest
Several attempts have been made to Doppler velocities. In particular, we angular resolution currently possible in
map the core of h Car and to peer into focused our efforts on mapping Brg and the near-infrared.
the structure of the wind-wind collision the He I 2s–2p lines.
region, and of the binary itself, at scales
of 5–10 astronomical units (au) or Images were recovered using SQUEEZE. References
2–4 mas. Long-baseline infrared interfer- Prior information necessary for the Baron, F. & Kloppenborg, B. 2010, Proc. SPIE, 7734,
ometry has been a d ­ ecisive technique ­reconstruction was included in both the 77344D
for such studies (van Boekel et al., 2003; spatial and spectral domains to obtain Buscher, D. F. 1994, Very High Angular Resolution
Weigelt et al., 2007). Astronomical simultaneous images of 35 different Imaging, IAU Symposium, 158, 91
Clementel, N. et al. 2015a, MNRAS, 450, 1388
­Multi-BEam combineR (AMBER) obser­ spectral channels, with a resolution as Clementel, N. et al. 2015b, MNRAS, 447, 2445
vations in 2014 allowed, for the first time, good as 1.75 mas (4 au; see Figure 1). GRAVITY Collaboration 2017, A&A, 602, A94
the recovery of aperture-synthesis Compared with the 2014 AMBER images, GRAVITY Collaboration 2018, A&A, 618, 125
images, at a resolution of ~ 6 mas of the the GRAVITY Brg ones revealed struc­- Mehner, A. et al. 2010, ApJ, 710, 729
Mehner, A. et al. 2012, ApJ, 751, 73
wind-wind collision cavity across Brg tural changes associated with the orbital Madura, T. I. et al. 2012, ApJ, 647, L18
(Weigelt et al., 2016). motion of the secondary. In particular, Madura, T. I. et al. 2013, MNRAS, 436, 3820
a bright “clump” is observed towards the Sanchez-Bermudez, J. et al. 2018, Experimental
southeast of the central core, which was Astronomy, 46, 457
Thiébaut, E. 2008, Proc. SPIE, 7013, 701311
Observing h Car with GRAVITY identified as part of shocked wind flowing van Boekel, R. et al. 2003, A&A, 410, L37
along the inner cavity walls after the last Weigelt, G. et al. 2007, A&A, 464, 87
The unique characteristics of h Car make hB periastron in 2014. Weigelt, G. et al. 2016, A&A, 594, A106
it a good candidate for increasing our
understanding of the role of multiplicity in The He I images revealed, for the first
shaping the fate of stars at the upper time, the distribution of this element in
end of the Initial Mass Function (IMF). It h Car’s core. We suggest that the ­partially

The Messenger 178 – Quarter 4 | 2019 33

GRAVITY Science DOI: 10.18727/0722-6691/5171

Precision Monitoring of Cool Evolved Stars:

Constraining Effects of Convection and Pulsation

Markus Wittkowski 1 can form in their atmospheres, and are and Alfvén waves (for example, Airapetian
Sara Bladh 2 subsequently expelled into the interstellar et al., 2010; Cranmer & Saar, 2011;
Andrea Chiavassa 3 medium via stellar winds. ­Yasuda & Kozasa, 2019; Rau et al., 2019).
Willem-Jan de Wit 1 Radiative pressure is currently being
Kjell Eriksson 2 Both AGB stars and RSGs are affected implemented in global CO5BOLD
Bernd Freytag 2 by pulsation and convection, but RSGs ­models. Magneto-hydrodynamical effects
Xavier Haubois1 show lower variability amplitudes than can, in principle, be described by
Susanne Höfner 2 AGB stars. For AGB stars, it has been ­CO5BOLD models (Freytag et al., 2012;
Kateryna Kravchenko 1 shown that pulsation and convection lead Steiner et al., 2014), but an application
Claudia Paladini 1 to strongly extended molecular atmos- to AGB and RSG stars requires further
Thibaut Paumard 4 pheres, where the temperature is low work.
Gioia Rau 5, 6 enough for dust condensation. Radiation
Peter R. Wood 7 pressure on dust then gives rise to a
­general mass outflow as the surrounding Pilot study with GRAVITY
gas is dragged along through friction
ESO (for example, Höfner & Olofsson, 2018). Time-series of interferometric observa-
Uppsala University, Department of tions provide the strongest tests of
Physics and Astronomy, Sweden For RSGs, it has been speculated that dynamical processes in the atmospheres
Université Côte d’Azur, Laboratoire the same processes may explain their of evolved stars, as they spatially resolve
Lagrange, Nice, France mass loss. However, Arroyo-Torres et al. the star and provide constraints on differ-
LESIA, Observatoire de Paris, Meudon, (2015) showed that current dynamic ent atmospheric layers, following the
France model atmospheres of RSGs, based on ­variability cycle of the star. However, such
NASA, Goddard Space Flight Center, pulsation and convection alone, cannot time-series are still very rare.
Greenbelt, USA explain the observed extensions of RSG
Catholic University of America, Depart- atmospheres, or how they can reach dis- Wittkowski et al. (2018) recently con-
ment of Physics, Washington, DC, USA tances where dust can form. This points ducted a pilot study measuring the varia-
Research School of Astronomy and to missing physical processes in current bility of the continuum radius and of
Astrophysics, ANU, Canberra, Australia RSG dynamic models. It translates into extended molecular layers for the oxy-
uncertainties in our general understand- gen-rich Mira star R Peg during science
ing of mass loss, as such processes may verification and early (P98) science
Mass loss from cool evolved stars is an to some degree also affect the atmos- ­operations, using the newly available
important ingredient of the cosmic mat- pheric structures of AGB stars and other near-infrared K-band beam combiner
ter cycle, enriching the Universe with cool giants. GRAVITY (GRAVITY Collaboration, 2017)
newly formed elements and dust. How- at the VLTI. This became possible
ever, physical processes that are not because of the improved performance of
considered in current models represent 1D and 3D model atmospheres the GRAVITY instrument compared to, for
uncertainties in our general under- example, the Astronomical Multi-BEam
standing of mass loss. Time-series of Significant advances are being made in combineR (AMBER), with an increased
interferometric data provide the strong- the development of dynamic atmosphere precision in visibilities, data for six base-
est tests of dynamical processes in the models of cool evolved stars. Latest lines in one snapshot, and a spectral
atmospheres of these stars. Here, we developments include 1D DARWIN (Bladh ­resolution of about 4000 across the full
present a pilot study of such measure- et al., 2019), and 3D CO5BOLD radiative K-band.
ments obtained with the GRAVITY hydrodynamics (RHD) simulations
instrument on the Very Large Telescope (Freytag et al., 2017; Höfner & Freytag, We showed that the continuum size and
Interferometer. 2019). In contrast to existing CO5BOLD the size in a bandpass that is dominated
and CODEX models, DARWIN models by water vapour were anti-correlated
include the wind acceleration region, with the visual light-curve. The size in the
Cool evolved stars which affects atmospheric structure and CO (2–0) line instead follows the visual
molecular features (Bladh et al., 2013, light-curve more closely, indicating a dif-
Asymptotic giant branch (AGB) and red 2015; Höfner et al., 2016), and may ferent — possibly more stable — behav-
supergiant (RSG) stars are located in account for some of the previously found iour of CO compared to water vapour
the Hertzsprung–Russell diagram at low discrepancies between AGB star models (Figure 1). The wavelength-dependent vis-
effective temperatures (about 2500– and interferometric observations. Addi- ibility variations could be reproduced by a
4500 K). They are major contributors to tional processes that may contribute set of CODEX (Ireland et al., 2008, 2011)
the integral luminosity of stellar systems, to larger atmospheric extension in RSG dynamic model atmospheres at phases
and they are major sources of the chemi- dynamic models include radiation pres- between 0.3 and 0.6. However, we
cal enrichment of galaxies. Owing to the sure on molecular lines (Josselin & Plez, noticed the following issues: (1) best-fit
low temperatures, molecules and dust 2007) or the effects of magnetic fields model phases did not correspond well

34 The Messenger 178 – Quarter 4 | 2019

Phase Figure 1. Variability of R Peg in the V-band (grey
crosses) and of the uniform disc angular diameter in
0.0 0.2 0.4 0.6 0.8
a near-continuum band (blue × symbols), and in
4 R Peg 2.25 µm (Cont.) bands dominated by H2O and CO (light blue and
14 pink × symbols, respectively). Also shown are sinu-
VLTI-GRAVITY 2.05 µm (H2O) soidal fits in the corresponding colours. The mini-
mum continuum size tracks the maximum light,
2.29 µm (CO 2-0)

Uniform disc diameter (mas)

6 which can be understood by the increase in effective
temperature while the star gets smaller in radius.
12 The minimum contribution of H2O also tracks the
V magnitude

maximum light, which relates to the destruction

8 of water vapour at maximum, and formation at mini-
mum, light. The contribution by CO is, however,
­largest at maximum light, indicating different, possi-
10 bly more stable, behaviour compared to H2O. From
10 Wittkowski et al. (2018).

12 8

7600 7700 7800
Time (JD-2 450 000)

with observed phases, and (2) the phere (Kravchenko et al., 2018, 2019). Ireland, M. et al. 2011, MNRAS, 418, 114
Kravchenko, K. et al. 2018, A&A, 610, A29
observed amplitude of the continuum Combined with spectro-interferometric
Kravchenko, K. et al. 2019, A&A, 632, A28
radius is 14% — this is smaller than GRAVITY observations on the VLTI, Rau, G. et al. 2019, ApJ, 882, 37
­predicted by CODEX model atmospheres the tomographic method will permit a Steiner, O. et al. 2014, PASJ, 66, S5
(45%–67%), and closer to those pre- simultaneous spectral and spatial char- Josselin, E. & Plez, B. 2007, A&A, 469, 671
Wittkowski, M. et al. 2018, A&A, 613, L7
dicted by 3D RHD simulations (Freytag acterisation of AGB and RSG star atmos-
Yasuda, Y. et al. 2019, ApJ, 879, 77
et al., 2017). The data covered only four pheres. By extracting interferometric
epochs, and more are needed to be visibilities at wavelengths contributing to
meaningfully compared to 3D models, different masks, we can measure the cor- Links
which show strong intra-cycle and cycle- responding geometrical extents of the 1
 RAVITY Science Verification: https://www.eso.
to-cycle irregularities. atmosphere and recover the link between org/sci/activities/vltsv/gravitysv.html
optical and geometrical depth scales. 2
GRAVITY consortium:
We plan to extend the GRAVITY pilot Based on observations made with the VLT Interfer-
study described above to a larger sample ometer at Paranal Observatory. We thank the
of cool evolved stars, and in particular ­GRAVITY Science Verification team1, the GRAVITY
to include a comparison of AGB stars, for consortium 2, the GRAVITY Collaboration (see
page 20), and the ESO science operation team for
which current models successfully pre- the development and operations of GRAVITY, and
dict observed extensions, and RSG stars, for their great support.
for which models and observations show
strong discrepancies in this respect. We
need a denser and wider phase sampling
compared to our plot study, including Airapetian, V. et al. 2010, ApJ, 723, 1210
intra-cycle and cycle-to-cycle variations, Arroyo-Torres, B. et al. 2015, A&A, 575, A50
to be able to make meaningful compari- Bladh, S. et al. 2013, A&A, 553, A20
Bladh, S. et al. 2015, A&A, 575, A105
sons to the latest dynamic models. Bladh, S. et al. 2019, A&A, 626, A100
Cranmer, S. R. & Saar, S. H. 2011, ApJ, 741, 54
We will be able to use more, and better- Freytag, B. et al. 2012, JCoPh, 231, 919
defined, atmospheric layers compared to Freytag, B. et al. 2017, A&A, 600, A137
GRAVITY Collaboration 2017, A&A, 602, A94
our pilot study by applying a tomographic Höfner, S. et al. 2016, A&A, 594, A108
method that relies on spectral masks Höfner, S. & Olofsson, H. 2018, A&ARv, 26, 1
selecting lines that form in given ranges Höfner, S. & Freytag, B. 2019, A&A, 623, A158
of optical depths in the stellar atmos- Ireland, M. et al. 2008, MNRAS, 391, 1994

The Messenger 178 – Quarter 4 | 2019 35

GRAVITY Science DOI: 10.18727/0722-6691/5172

Multiple Star Systems in the Orion Nebula

GRAVITY Collaboration (see page 20) High angular resolution observations are and a semi-major axis of 18.2 ± 0.3 au.
crucial to pinning down the dominant Additionally, we determined a new Orbit
mode of massive star formation. One of for q1 Ori D2, with a semi-major axis of
GRAVITY observations reveal that most the closest massive star forming regions 0.77 ± 0.03 au and a period of 53.05 ±
massive stars in the Orion Trapezium is the Orion Nebula Cluster, located at a 0.06 days.
cluster live in multiple systems. Our distance of 414 ± 7 pc (for example,
deep, milliarcsecond-resolution interfer- Reid et al., 2014). As such the Orion Neb-
ometry fills the gap at 1–100 astrono­ ula has been the target of many previous Most massive stars live in multiple
mical units (au), which is not accessible observations. The superb angular resolu- systems
to traditional imaging and spectros- tion and sensitivity of GRAVITY using
copy, but is crucial to uncovering the the VLTI can reveal details on the crucial Massive stars are more often found in
mystery of high-mass star formation. scales of 1–100 au, which had remained multiple systems than are lower mass
The new observations find a signifi- mostly unexplored until now. stars. For example, Duchene & Kraus
cantly higher companion fraction than (2013) and Sana et al. (2014) found
earlier studies of mostly OB associa- increasing numbers of stars in compan-
tions. The observed distribution of Observations with GRAVITY ion systems with higher stellar mass.
mass ratios declines steeply with mass Additionally, the average number of com-
and follows a Salpeter power-law We observed the 16 brightest, most mas- panion stars increases with higher mass.
initial mass function. The observations sive stars in the Orion Nebula Cluster, Our observations confirm this trend
therefore exclude stellar mergers as with masses between 2 and 44 M⊙. The and our results are comparable to those
the dominant formation mechanism for observations were mostly done with of Sana et al. (2014). Orion’s O-type stars
massive stars in Orion. the Auxiliary Telescopes in astrometric have an average of 2.3 ± 0.3 companions.
configuration. Data were reduced with
the standard GRAVITY pipeline (GRAVITY Plotting the number of all our observed
The formation of massive stars Collaboration, 2017). The interferometric stars and their companions against stellar
data were then fitted to a binary star mass, we find the mass function well
The formation of massive stars remains a model, providing the flux ratio of the com- described by a power law with an expo-
mystery. Hidden in their parental gas panion to the main star, and the separa- nent of G = 1.3 ± 0.3 (Figure 2). This
and dust clouds, it is unclear how their tion vector between the two components matches the initial mass function (IMF)
seeds can accrete so much matter before (GRAVITY Collaboration, 2018). for field stars (see, for example, Salpeter,
the repulsive forces from thermal pres- 1955).
sure and radiation prevent the formation We focused first on the central region,
of a protostar. The most discussed the Orion Trapezium Cluster, home of Ori- To constrain star formation scenarios,
­scenarios are competitive accretion and on’s most massive, visible star, q1 Ori C. we compare predictions to our observa-
core accretion (see, for example, Tan et The 16 observed objects have a total tions. For both core accretion and com-
al., 2014 and references therein). Another of 22 companions; see Figure 1 for an petitive accretion, the number of stars
possibility is the ­collision of two stars, overview. in companion systems and the number of
merging into a more massive star. companions should rise with mass
With GRAVITY, we found three previ- (Clarke, 2001). Therefore, both scenarios
Core accretion is a scaled-up version of ously unknown companions and we con- would match our observations. The situa-
standard star formation applicable to firm a suspected companion for n Ori tion is different for the correlation
stars similar to our Sun. In this scenario it (­Grellmann et al., 2013). The newly dis- between the companion masses. While
is a single core that accretes its mass covered stars belong to the systems of competitive accretion shows no clear
independently of other sibling cores. The q1 Ori B, q2 Ori B, and q2 Ori C. We deter- correlation between the primary and sec-
mass of the star is then set at the begin- mined their separation, and from the ondary mass, with a mass distribution
ning of the process, determined by the flux ratio we could estimate the masses that could follow a Salpeter IMF (for
available mass in the accretion volume. of all new companions (see GRAVITY example, Tan et al., 2014) or a top-heavy
An alternative explanation is formation by Collaboration, 2018 for more details). companion mass distribution (Bate,
competitive accretion (for example, Tan q1 Ori B is a system of particular interest, ­Bonnell & Bromm, 2002), core accretion
et al., 2014), where several cores com- as it consists of six objects in total. These results in a strong correlation between
pete for the available mass, culminating in objects are all gravitationally bound, the companion masses, which we do not
hierarchical systems with stars of differ- though it is suspected that the system is observe. Also, the companion separation
ent masses. Unlike in the core-accretion only temporarily stable (Close et al., should correlate with system mass for
model, their masses are not pre-defined, 2013). q1 Ori C is accompanied by two core accretion. For competitive accretion,
but depend on the interaction with each companion stars, one spectroscopic the separation should inversely correlate
other. A third possibility for the formation companion and one known companion with system mass (Bonnell & Bate, 2005).
of massive stars is stellar mergers, where with a determined orbit. With GRAVITY In Orion, we observe no correlation
two colliding stars end up in a more mas- observations, we could refine the orbit of between separation and system mass
sive object. q1 Ori C2 to have a period of 11.4 ± 0.2 yr (Figure 3), which is inconsistent with

36 The Messenger 178 – Quarter 4 | 2019

Figure 1. Overview of all observed
multiple stars in the Orion Nebula.
1 au 100 au
0.1 au 100 au
The observed 16 systems comprise a
B5 B1 B4 total of 22 companions. The scale of
the separation of the companion is
indicated in the figure. The coloured
B1, 5 images of q1 Ori B are from observa-
0.1 au
tional data, except the greyscale B1,
B1, 5, 6 B5 system, which is only representa-
NU Ori tive. The image of B1, 5 and B6 is a
B6 5 au reconstructed image of GRAVITY
0.3 au observations. The orbital positions,
B3 which are indicated for q1 Ori C and
q1 Ori D, are the positions given in
GRAVITY Collaboration (2018) and
0.1 au 157 au
θ2 Ori A ­p revious literature. The other greyscale
0.9 au close-up images are for illustrative
purposes only. This figure is taken
from GRAVITY Collaboration (2018).
16 au

E 100 au
θ2 Ori C
θ2 Ori B C
40 au 580 au
F 0.7 au

10 au

HD 37115 Figure 3 (below). Companion separa-

0.5 au tion for all of Orion’s multiple star sys-
TCC59 tems, sorted by mass of the primary
100 au 0.4 au 56 au star. Each system is indicated by a dif-
ferent colour. The dot size scales
with the square root of the companion
mass. There are as many companions
in the range 0.1–1 au as in the range
1–100 au. The dashed circles around
companions of q2 Ori C and TCC 59
indicate missing information about the
masses. (GRAVITY Collaboration, 2018)

IMF, Γ = 1.3
θ2 Ori A (39.0 ± 14.0 M๬ )
IMF, Γ = 1.6
10 –1 IMF, Γ = 1.0 θ1 Ori C (33.0 ± 5.0 M๬ )
Number of stars

Nu Ori (16.01 ± 3 M๬ )
System & primary mass (M๬ )

θ1 Ori D (16.0 ± 1.0 M๬ )

θ2 Ori B (14.8 ± 3.4 M๬ )

θ1 Ori A (14.0 ± 5.0 M๬ )

10 –2
θ1 Ori B (7.2 ± 0.2 M๬ )

HD 37115 (5.4 ± 0.4 M๬ )

θ2 Ori C (4 ± 1.0 M๬ )
Mass (M๬ )
θ Ori E (2.81 ± 0.05 M๬ )

Figure 2. The observed mass distribution of the TCC 59 (2 ± 0.5 M๬ )

observed stars and their companions as a
­normalised histogram, and the initial mass functions
as proposed by, for example, Kroupa (2001). 10 –1 10 0 101 10 2
Companion separation (au)

The Messenger 178 – Quarter 4 | 2019 37


either scenario. In addition, competitive stars live in multiple systems. We do not Acknowledgements
accretion predicts an anti-correlation see a strong preference for either core
See page 23.
between the mass ratio of the companion collapse or competitive accretion among
to primary star and their separation, the massive stars of Orion. The Salpeter
which we do not see in our data. If stellar IMF hints towards competitive accretion, References
collisions were the dominant formation whereas the lack of correlations between
Bate, M. R., Bonnell, I. A. & Bromm, V. 2002,
process, we would expect a strong devi- separation, system mass, primary and MNRAS, 336, 705
ation from the Salpeter IMF (Moeckel & companion masses contradicts it. We Bonnell, I. A. & Bate, M. R. 2005, MNRAS, 362, 915
Clarke, 2011). Thus we can exclude stellar can exclude the collision of stars as the Clarke, C. J. 2001, The Formation of Binary Stars,
IAU Symposium, 200, 346
mergers as the dominant formation main mechanism for the formation of high
Close, L. M. et al. 2013, ApJ, 774, 13
mechanism for massive stars in Orion. mass stars in Orion, which would result Duchêne, G. & Kraus, A. 2013, ARA&A, 51, 269
in a strong deviation from the Salpeter GRAVITY Collaboration 2017, A&A, 602, A94
IMF. Our GRAVITY results highlight the GRAVITY Collaboration 2018, A&A, 620, A116
Grellmann, R. et al. 2013, A&A, 550, 531
Summary & conclusions crucial role of interferometry in filling the
Kroupa, P. 2001, MNRAS, 322, 231
gap between 1 and 100 au, which is Moeckel, N. & Clarke, C. J. 2011, MNRAS, 410, 2799
We probed the Orion Nebula for massive not accessible with traditional imaging Reid, M. J. et al. 2014, ApJ, 783, 130
multiple star systems with separations and spectroscopic techniques. Salpeter, E. E. 1955, ApJ, 121, 161
Tan, J. C. et al. 2014, Protostars and Planets VI, ed.
between 1 and 100 au. Almost all massive
Beuther, H. et al., (Tucson: Univ. of Arizona), 149

DOI: 10.18727/0722-6691/5173

Probing the Discs of Herbig Ae/Be Stars at

Terrestrial Orbits

GRAVITY Collaboration (see page 20) are born in and/or migrate into the inner- gas might accrete onto the star through
most regions close to the host star. As magnetospheric accretion or be launched
discs evolve, different phenomena such through winds and jets, where dust is
More than 4000 exoplanets are known as photoevaporation, mass-loss through thermally processed, sublimated, and
to date in systems that differ greatly winds and jets, and dynamical clearing from where it can be redistributed into
from our Solar System. In particular, by newly-formed planets will disperse the the outer disc. Identifying dust traps
inner exoplanets tend to follow orbits disc material. Thus disc evolution and and other planetary signposts such as
around their parent star that are planet formation are linked processes. dynamical perturbations in the disc is
much more compact than that of Earth. Observing the inner regions with suffi- an important goal if we are to constrain
These systems are also extremely cient angular resolution is crucial for bet- inner planet formation mechanisms.
diverse, covering a range of intrinsic ter understanding the key physical pro-
properties. Studying the main physi- cesses at play and how they combine to
cal processes at play in the innermost lead to the formation of an exoplanetary The diverse nature of the inner discs
regions of the protoplanetary discs system.
is crucial to understanding how these In this contribution, we highlight GRAVITY
planets form and migrate so close to Thanks to high angular resolution imaging observations that reveal the morphology
their host. With GRAVITY, we focused in the optical range with the Spectro-­ of the inner dusty discs. The near infrared
on the study of near-infrared emission Polarimetric High-contrast Exoplanet emission detected with GRAVITY 1 and
of a sample of young intermediate- REsearch instrument (SPHERE; Beuzit et the Precision Integrated Optics Near-
mass stars, the Herbig Ae/Be stars. al., 2019), and at (sub-)millimetre wave- infrared Imaging ExpeRiment (PIONIER 2;
lengths with the Atacama Large Millime- Le Bouquin et al., 2011) arises mostly in
ter/submillimeter Array (ALMA partnership the dust sublimation front of the inner part
Dust in the innermost regions of the et al., 2015), rings, gaps, spiral arms, of the protoplanetary disc. We observe
young intermediate-mass stars warps, and shadows have been revealed wedge-shaped rims, with a smooth radial
in the outer disc on scales ranging from distribution of dust that is much wider
The formation and evolution of proto­ a few tens to a few hundreds of astro- than would be expected for a single dust
planetary discs are important stages in nomical units (au). GRAVITY uniquely component (GRAVITY Collaboration et
the lifetimes of stars. Terrestrial planets probes the innermost few au where hot al., 2019). We suggest that these inner-

38 The Messenger 178 – Quarter 4 | 2019

observed in the low-luminosity, less
HD 100546 (g) Gapped discs (g) HD 100546 (g) ­massive members of our sample that
100.0 are older than 1 Myr.

HD 142666 (g) HD 169142 (g) HD 169142 (g)

Evolution of the inner structure

Closest to the ZAMS

HD 97048 (g) HD 97048 (g)
N-to-K size ratio

HD 38120 HD 38120 An underlying question related to disc

classification is whether the sources
HD 144432 (g)
HD 150193
HD 259431
HD 150193
with flared/gapped and flat/continuous
HD 142527 (g)
HD 158643 discs form an evolutionary sequence.
HD 179218 (g) ­Dullemond and Dominik (2004) proposed
HD 139614 (g) HD 95881
HD 163296
HD 98922
HD 163296 that discs might start out with a flared
HD 45677
HD 135344 (g)
HD 144668 HD 144668 shape, then become flat when going
Continuous discs
through the process of grain growth. On
the other hand, Maaskant et al. (2013)
1.0 ­proposed that both can evolve from the
1.0 2.0 5.0 10.0
Star mass (M๬ )
primordial flared discs. More recently,
Menu et al. (2015) suggested that either
Figure 1. N-to-K size ratio of the discs as a function tral Energy Distributions (SEDs; Meeus et each group could follow a distinct evolu-
of the mass of the central stars. The blue diamonds
al, 2001); i.e., sources with decreasing tionary path from continuous to gapped
denote flared/gapped discs (as sketched in the
top inset), while the red triangles denote flat/continu- mid-infrared SEDs exhibit a flat geometry disc, or flat gapped discs could later
ous discs (as sketched in the bottom inset) accord- while those with flat or rising mid-infrared evolve into flared discs with larger gaps.
ing to the Meeus (2001) classification. The square SEDs have flared discs. For these latter
symbol denotes an unclassified star. The gapped
SED shapes, Maaskant et al. (2013) pro- We use the masses and ages provided
sources are identified as (g). The green area spots
the 2 M⊙ objects and the arrow indicates the position posed that they could indicate a gapped by ESA Gaia observations (Vioque et al.,
with respect to the Zero-Age Main Sequence (ZAMS). disc structure. 2018) and focus on the relative ages of
objects with similar masses to avoid
We compare the disc sizes in the ­well-known mass bias effects for the age
most regions host grains of different sizes N-band, derived from MIDI (Menu et al., ­estimation of the young stellar objects.
and/or compositions so that some can 2015), with our K-band measurements. For 2 M⊙ objects, we observe a transition
survive near the inner rim while others are To first order, the K-band and N-band from flat/continuous to flared/gapped
further away. Moreover, GRAVITY reveals sizes increase proportionally. Using shapes, and an increase of the N-to-K-
a slight asymmetry in most of these the N- to K-band size ratios as a proxy, band size ratio when the relative age
discs that could be explained by inclina- we look for general trends for about increases (Figure 1). However, owing to
tion effects for more than half the objects, 20 objects in our GRAVITY sample; the limited size of our GRAVITY sample,
while an intrinsic asymmetry should be gapped sources exhibit a large N-to-K- we cannot establish any clear universal
invoked for others. From the observa- band size ratio, and large ratios are only evolution mechanism across the Herbig
tions of 27 targets, we confirm the size-­
luminosity­relationship. For the luminous
stars (around 103 L⊙ ), a large scatter us
around the mean relation is observed, 10.0 ic al r
HD 45677 Crit Figure 2. K-band emis-
pointing towards a range of compositions sion location as meas-
of the inner dusty discs. ured by GRAVITY as a
K-band emission location (au)

function of star mass.

The dashed red line cor-
responds to the critical
Flat, flared, gapped and continuous radius where the gap
discs is expected to form as a
1.0 result of EUV/FUV/X-ray
heating from the central
To trace the disc regions beyond the dust star. This photoevapo­
sublimation rim, mid-infrared interfer­ ration phenomenon
ometry is a powerful tool, as it probes leads to a fast depletion
EUV/X-ray of the inner disc and
dust at temperatures down to ∼ 300 K. a void central cavity (as
After the MID-infrared Interferometric sketched in the top
instrument (MIDI), the Multi AperTure inset) while young plan-
0.1 ets and EUV/X-ray
mid-Infrared SpectroScopic Experiment
­p hotoevaporation will
(MATISSE 3 ) can now investigate disc open gaps and not
­flaring at tens of au and can help to ques- 1.0 2.0 5.0 10.0 20 deplete the inner disc
tion disc classifications based on Spec- Star mass (M
M๬ ) quickly (bottom inset).

The Messenger 178 – Quarter 4 | 2019 39


Ae/Be mass range and need additional emission we measured with GRAVITY is GRAVITY Collaboration et al. 2019, A&A, 632, A53
Le Bouquin, J.-B. et al. 2011, A&A, 935, A67
observations. located at positions smaller than the
Lopez, B. et al. 2018, SPIE, 10701, 107010Z
­critical radius where the gap is expected Maaskant, K. M. et al. 2013, A&A, 555, A64
to form as a result of to extreme-/far-­ Meeus, G. et al. 2001, A&A, 365, 476
Gap formation scenarios ultraviolet/X-ray heating, the discs in our Menu, J. et al. 2015, A&A, 581, A107
Vioque, M. et al. 2018, A&A, 620, A128
sample might be shaped by forming
Zhang, S. et al. 2018, ApJ, 869, L47
Gaps in protoplanetary discs can be young planets rather than by depletion
found in concentric arrangements from resulting from photoevaporation (Figure 2).
the inner regions out to large distances Notes
— as nicely evidenced by ALMA images With PIONIER, GRAVITY and MATISSE, 1
 RAVITY operates in the near-infrared K-band, i.e.,
(Zhang et al., 2018). Clearing by dynami- the VLTI is perfectly equipped to reveal with wavelengths between 2 and 2.5 µm.
cal effects due to newly-born planets the gas and dust distributions in proto­ 2
PIONIER at the VLTI operates in the near-infrared
and photoevaporation by extreme- planetary discs at unprecedented angular H-band, i.e., with wavelengths between 1.5 and
1.8 µm.
and far-ultraviolet (EUV/FUV) and X-ray and spectral resolution. 3
M ATISSE at the VLTI operates in the mid-infrared
radiation from the central star are key L-, M-, and N-bands, i.e., with wavelengths
processes of disc dispersal through gap between 3 and 13 µm (Lopez et al., 2018).
and inner cavity formation. In the photo- References
evaporation scenario, gap formation ALMA Partnership et al. 2015, ApJ, 808, L1
takes a few 10 6 years and inner disc Beuzit, J.-L. et al. 2019, A&A, 631, A155
depletion takes about 10 5 years (Gorti et Gorti, U. et al. 2009, ApJ, 705, 1237
al., 2009). Since almost all the K-band Dullemond, C. & Dominik, C. 2004, A&A, 421, 1075

DOI: 10.18727/0722-6691/5174

Spatially Resolving the Inner Gaseous Disc of the

Herbig Star 51 Oph through its CO Ro-vibration Emission

GRAVITY Collaboration (see page 20) although there are many studies of the emission at 2.3 microns is a good tracer
outer part of the disc, there are very few of the hot inner gaseous disc. Therefore,
on the inner disc, in particular the inner spatially resolved observations of the CO
Near-infrared interferometry gives us gaseous disc. This hinders our under- ro-vibrational transitions are crucial to
the opportunity to spatially resolve standing of the physical processes taking constraining the dynamics and chemical
the circumstellar environment of young place in this inner part of the disc. In composition of the inner dust-free disc.
stars at sub-astronomical-unit (au) addition to the study of the size and
scales, which a standalone telescope shape of the continuum emission origi-
could not reach. In particular, the sensi- nating in protoplanetary discs around The source
tivity of GRAVITY on the VLTI allows young stellar objects (YSOs), GRAVITY’s
us to spatially resolve the CO overtone spectral resolution of up to R = 4000 51 Oph is a fast-rotating star (v sin i =
emission at 2.3 microns. In this article, facilitates studies of the gaseous compo- 267 km s –1; Dunkin, Barlow & Ryan, 1997)
we present a new method of using nent of circumstellar material. The two and is located at a distance of 123 par-
the model of the CO spectrum to recon- most prominent components are hydro- secs (Lindergren et al., 2016; GAIA Col-
struct the differential phase signal gen, in the form of the Brackett g and laboration et al., 2018). Its spectrum is full
and extract the geometry and size of higher levels of the Pfund recombination of atomic and molecular lines and it is
the emitting region. lines, and molecular gas as traced by CO one of the very few Herbig Ae/Be stars
ro-vibrational transitions. A direct tracer that shows bright 2.3-micron CO over-
of the gas in the disc is the CO molecule. tone emission, making it an ideal candi-
Protoplanetary discs at high angular The CO emission is present at different date for near-infrared interferometric
resolution scales throughout the disc: from the studies (Thi et al., 2005; Berthoud et al.,
outer, cooler regions detected at millime- 2007; Tatulli et al., 2008).
Circumstellar discs are crucial to under- tre wavelengths to the inner, warmer
standing how stars and planets form. regions detected at near-infrared wave- The CO spectrum of the star has been
They contain both gas and dust and, lengths. In particular, the CO ro-vibrational extensively studied spectroscopically

40 The Messenger 178 – Quarter 4 | 2019

Figure 1. Spectrum (top), and differential phases regions. The line emitting regions origi-
of the GRAVITY interferometric data at one epoch.
nate closer to the star than the contin-
Each colour represents a baseline.
2–0 3–1 4–2 5–3
uum. There is a strong differential phase
signal (bottom panels of Figure 1) indi­
and indicates gas close to the star show- cating a photocentre shift of the line with
1.15 ing Keplerian rotation (Thi et al., 2005; respect to the continuum. There is no
Berthoud et al., 2007). Near-infrared indication of an asymmetric circumstellar

1.10 interferometric observations using the environment.

Astronomical Multi-BEam combineR
(AMBER) — one of the first-generation By fitting our interferometric observations
interferometric instruments on the VLTI — (i.e., spectrum, visibilities and differential
confirmed that the CO is emitted from the phases) with a local thermodynamic
1.00 very inner disc regions, within the dust ­equilibrium (LTE) model of the CO, we
sublimation radius (Tatulli et al., 2008). find that the CO is emitted from a rela-
Unfortunately, these observations were at tively warm (2400 K) and dense (NCO ~
15 102 m, 46° low spectral resolution (R = 1500) and 2 × 1021 cm – 2) region, consistent with gas
56 m, 163° no differential phase signal was retrieved. rotating in a ring at Keplerian velocity
135 m, 43°
(v sin i ~ 147 km s –1) and located at
roughly 0.1 au from the star (more details
Differential phase

The circumstellar environment of in Koutoulaki et al., in preparation; see

5 51 Oph also the spatial scales probed in Fig-
ure 3). From the model, an intensity map
51 Oph was observed with GRAVITY could be created since we know the
­during Guaranteed Time Observations intensity at each azimuthal angle. An
(GTO) at two epochs using the 1.8-metre example is shown in Figure 2 for three
–5 Auxiliary Telescopes (ATs). From an opti- velocity channels at the blueshifted
cal interferometer like the VLTI we can (left panel), peak (middle panel), and red-
usually extract the following information: shifted (right panel) parts of the first
spectra, visibilities, differential phases band head. This intensity map can be
15 91 m, 79° and closure phases. The visibility gives an used as an input to the latest version of
34 m, 34° estimate of the size of the emitting region, the interferometric software “Astronomi-
117 m, 67°
10 with a visibility equal to 0 meaning a cal Software to PRepare Observations”
resolved source, while 1 is a point source. — ASPRO2 — to simulate GRAVITY
Differential phase

The differential phase is the photocentre observations. By varying the inclination

shift of the line with respect to the con­ and position angle the software creates
tinuum and the closure phase is an indi- synthetic observations that can be com-
0 cator of asymmetries in the circumstellar pared with the real ones.
–5 Assuming Keplerian rotation, the combi-
One epoch of our VLTI/GRAVITY obser- nation of the v sin i measurement, deter-
vations of 51 Oph is shown in Figure 1. mination of the inclination, i, of the CO
The spectrum shows four bright band
heads (u = 2–0, 3–1, 4–2 and 5–3). The Figure 2. Intensity maps of the blueshifted (left),
2.30 2.35 2.40
Wavelength (µm) environment of 51 Oph is compact, both peak (middle), and redshifted (right) parts of the first
in the line and the continuum emitting band head.

1.0 1.0 1.0

0.10 0.10 0.10

0.8 0.8 0.8
Normalised intensity

0.05 0.05 0.05

0.6 0.6 0.6
Size (au)

0.00 0.00 0.00

0.4 0.4 0.4
– 0.05 – 0.05 – 0.05
0.2 0.2 0.2
– 0.10 – 0.10 – 0.10

0.0 0.0 0.0

0.10 0.05 0.00 – 0.05 – 0.10 0.10 0.05 0.00 – 0.05 – 0.10 0.10 0.05 0.00 – 0.05 – 0.10
Size (au) Size (au) Size (au)

The Messenger 178 – Quarter 4 | 2019 41

GRAVITY Science GRAVITY Collaboration, Inner Gaseous Disc of 51 Oph

Figure 3. Cartoon of
a protoplanetary disc
based on Testi et al.
2400 K, GRAVITY, 1500 K, GRAVITY, SPHERE, NIR (2014). The different
boxes correspond to
CO bandheads NIR continuum scattered light different spatial scales
and temperatures of
the disc as observed by
different instruments.

300–900 K, MATISSE, 20 K, ALMA, (sub-mm)

8500 K, star
MIR continuum continuum

0.1 au 1 au 10 au 100 au

ring, the angular extent of the CO, and ferometric observables can provide new References
the known distance to 51 Oph from GAIA insights into the geometry and size of the
Berthoud, M. G. et al. 2007, ApJ, 660, 461
Data Release 2, results in a direct meas- gaseous disc very close to the star. In Dunkin, S. K., Barlow, M. J. & Ryan, S. G. 1997,
urement of the mass of the central star of the case of 51 Oph we have been able, MNRAS, 286, 604
3.9 ± 0.6 M⊙ . for the first time, to observationally Gaia Collaboration et al. 2018, A&A, 616, A1
Lindegren, L. et al. 2016, A&A, 595, A4
constrain the physical properties of the
Muzerolle, J. et al. 2004, ApJ, 617, 406
GRAVITY has opened a new window gas at 0.1 au from the star; we find Tatulli, E. et al. 2008, A&A, 464, 55
enabling the use of molecular lines to ­physical properties consistent with those Testi, L. et al. 2014, Protostars and Planets VI,
probe the circumstellar environment of expected from LTE models of the ed. Beuther, H. et al., (Tucson: University of
­A rizona Press), 339
young stars. This new technique of gas content of the disc (Muzerolle et al.,
Thi, W. F. et al. 2005, A&A, 430, L61
­combining the spectrum fit with inter­ 2004).
ESO/P. Carrillo

The 8.2-metre Unit Tele-

scopes of the Paranal
Observatory in silhouette
against the Sun.

42 The Messenger 178 – Quarter 4 | 2019

GRAVITY Science DOI: 10.18727/0722-6691/5175

Spatially Resolving the Innermost Regions of the

Accretion Discs of Young, Low-Mass Stars with GRAVITY

Claire L. Davies 1 ­ ircumsecondary discs. During GRAVITY

c expected location of the dust sublimation
Edward Hone 1 science verification, we used the VLTI’s rim and observations with GRAVITY on
Jacques Kluska 2 Auxiliary Telescopes (ATs) to observe longer baselines are required to confirm
Alexander Kreplin1 the individual components of the CO Ori this result.
young stellar binary (Programme ID 60.A-
9159; PI Davies). While the primary com- CO Ori A also exhibits a 12.4-year
 strophysics Group, University of
A ponent (CO Ori A) was sufficiently bright ­peri­odicity in its optical photometry
­E xeter, UK (K = 6.0 magnitudes) for standard single ­(Rostopchina et al., 2007), potentially
Instituut voor Sterrenkunde (IvS), field mode observations, the secondary indicating the presence of an additional,
KU Leuven, Belgium (CO Ori B: K = 9.0 magnitudes) required as yet undetected, companion. Using
GRAVITY’s unique dual-field mode. Our our best-fit geometric modelling result
observations confirmed the existence of and the fringe tracker visibilities and clo-
Low-mass, young stars — the T Tauri individual circumstellar discs around both sure phases obtained for CO Ori A, we
stars — make up the majority of young components and spatially resolved them searched for evidence of off-centre
stellar objects. They have been for the first time. brightness contributions which may indi-
­relatively unexplored with optical long cate the presence of such a companion.
baseline interferometry owing to the CO Ori B displayed Brg emission in its Within the field of view probed by our
cooler temperatures of their stellar spectrum, which is typically associated short baseline observations, we found
­photospheres which makes them fainter with accretion and related outflow pro- no evidence for an additional companion
and more compact than the more cesses in young stars. This provided to CO Ori A. Furthermore, we were able
­frequently studied intermediate mass, ­further evidence for the existence of a cir- to rule out the presence of companions
young stars — the Herbig Ae/Be cumsecondary accretion disc. Mean- providing as much as 3.6 per cent of the
stars. With its greater flux sensitivity, while, the absence of Brg emission in the total K-band flux within 7.3 to 20 milli-
­GRAVITY has allowed us to explore spectrum of CO Ori A may indicate that arcseconds (Figure 1). These results are
T Tauri stars at high angular resolution accretion onto the primary star is weaker presented in greater detail in Davies et al.
in unprecedented detail. Here we pres- or otherwise inhibited. We investigated (2018).
ent highlights from two such studies. whether the characteristic size of the
near-infrared emission is consistent with
the location of the dust sublimation rim, Magnetospherically truncated discs
GRAVITY is enabling substantial progress as has been seen to be the case for
in high-angular-resolution studies of discs around more massive young stars In 2018, we used GRAVITY’s high spectral
star formation. Its spectrograph’s greatly (for example, L
­ azareff et al., 2017). We dispersion (R = 4000) mode to observe
improved K-band flux sensitivity compared compared the continuum visibilities to the low-mass, young star CW Tau (Pro-
to the first-generation VLTI instrument geometric models incorporating a central gramme ID 102.C-0755; PI Hone) in single-
Astronomical Multi-BEam combineR point source and a Gaussian component, field mode. We fit the fringe tracker visi-
(AMBER), for example, has unlocked the finding a characteristic radius of 2.31 ± bilities with a two-­dimensional geometric
previously poorly studied sample space 0.04 milliarcseconds. At first glance, this model comprising a central point source
occupied by the low-mass (T Tauri) stars appears large compared with the (simulating the star), a ring (simulating
— which comprise the majority of young
stellar objects. The flux sensitivity is

­further increased by GRAVITY’s unique
dual-field mode which allows for longer

integration observations with the spectro-  
graph if a sufficiently bright, neighbour-
ing star exists within a few arcseconds. 
  Figure 1. CO Ori A sen-
GRAVITY’s ability to observe with high

sitivity map produced


angular and high spectral resolution  using the best-fit model

across the K-band allows us to study the   from our modelling
of the continuum visibili-
dynamic inner regions of protoplanetary 
ties. Each pixel in the
discs and to directly measure the pro-   map is coloured to
cesses by which mass is accreted onto l reflect the maximum
stars and launched into outflows. possible flux contribu-
  tion that a companion
l could have at that posi-
tion and remain unde-
Discs identified around the low-bright- l   tected. As our observa-
ness stars in the binary CO Ori tions do not sample
the entire uv plane, we
l    are not sensitive to
Wide-separation young stellar binary sys-      l l l l  ­c ompanions in regions
tems often feature circumprimary and 61HFGS@RBDMRHNML@R outlined in red.

The Messenger 178 – Quarter 4 | 2019 43

GRAVITY Science Davies C. L. et al., Accretion Discs of Young, Low-Mass Stars

PA = 120.5° ± 11.0° 150

0.04 Continuum
5 PA = –77.3 BL = 59.8 PA = 39.5 BL = 40.7 PA = 46.5 BL = 78.2 Brγ line 100

Δ Declination (mas)

Velocity (km s –1)

0 50
DP (degree)

–5 0.00 0
5 PA = 54.2 BL = 37.1 PA = 70.85 BL = 120.0 PA = 84.45 BL = 88.8
– 50
0 – 0.02
2.164 2.166 2.168 2.164 2.166 2.168 2.164 2.166 2.168 – 0.04 –150
Wavelength (µm)

0.04 0.02 0.00 – 0.02 – 0.04

Figure 2. Differential phases (left panel) and corre- Δ Right ascension (mas)
sponding photocentre displacement vectors (right
panel) across the Brg line for CW Tau. The black lines
in the left panel indicate how well the photocentre shifted and blueshifted material along telescope beam combination has also
shifts match the differential phases. The solid black a position angle of 120.5 ± 11.0 degrees, enabled more efficient uv-plane sampling.
lines in the right panel show the best-fit position (Figure 2). The position angle of the The arrival of NAOMI, the adaptive optics
angle of the motion between redshifted and
motion traced by the photocentre shifts system for the ATs, during the summer of
blueshifted vectors, with the dashed line indicating
the uncertainty. is closer to the minor axis of the disc of 2019 has further improved the flux sensi-
CW Tau as seen by ALMA (150.7 degrees; tivity and we look forward to f­urther prob-
Bacciotti et al. 2018), suggesting that ing how well our understanding of star
disc emission on milliarcsecond scales) the Brg emission predominantly traces formation, garnered from the study of
and an extended Gaussian component motion out of the disc plane. This indi- intermediate-mass young stars, is trans-
(­simulating over-­resolved larger-scale disc cates the existence of a complex velocity ferable to the low-mass T Tauri stars.
emission). The best-fit ring radius from field traced by the Brg emitting gas, such
our model (0.56 ± 0.2 milliarcseconds) as the launching of a jet or the motion
corresponds to a physical distance of of material being carried along magneto- Acknowledgements
0.074 ± 0.03 astronomical units. We find spheric accretion funnels. Indeed, the We acknowledge support from ERC Starting
this distance to be consistent with the presence of a jet emerging away from the Grant “ImagePlanetFormDiscs” (Grant Agreement
magnetospheric truncation radius and, star-disc system, towards the south-east, No. 639889) and help from ESO staff astronomers
from this, estimate a dipolar magnetic along a position angle of ~ 150 degrees and the GRAVITY consortium and Science Verifi­
cation team in the execution and reduction of our
field component strength of ~ 2 kG for has been observed previously by, for observations.
CW Tau. This value is consistent with example, Gomez de Castro (1993) and
those previously found for low-mass McGroarty, Ray & Froebrich (2007).
young stars via spectropolarimetric tech- These results form part of the PhD thesis References
niques (for example, Donati et al., 2010; of Edward Hone and will be presented in Bacciotti, F. et al. 2018, ApJL, 865, 12
Hill et al., 2019). more detail in Hone et al. (in preparation). Davies, C. L. et al. 2018, MNRAS, 474, 5406
Donati, J.-F. et al. 2010, MNRAS, 409, 1347
We also used our differential phases to Gomez de Castro, A. I. 1993, ApJL, 412, 43
Hill, C. A. et al. 2019, MNRAS, 484, 5810
calculate model-independent photocen- GRAVITY’s impact and outlook for the Lazareff, B. et al. 2017, A&A, 599, 85
tre shifts across the Brg emission line, future McGroarty, F., Ray, T. P. & Froebrich, D. 2007, A&A,
tracing the small-scale displacement of 467, 1197
the centroid as a function of wavelength. GRAVITY has extended high-resolution Rostopchina, A. N. et al. 2007, Astron. Rep., 51, 55
The photocentre displacement vectors studies of star formation to the low-
reveal a clear displacement between red- mass T Tauri stars. The availability of four-­
ESO/S. Otalora

The landscape of the

Chajnantor Plateau on
which the antennas of
the Atacama Large
­Milli­m eter/submillimeter
Array (ALMA) are

44 The Messenger 178 – Quarter 4 | 2019

GRAVITY Science DOI: 10.18727/0722-6691/5176

When the Stars Align — the First Resolved

Microlensed Images

Subo Dong 1 ­ instein reluctantly published the idea of

E effect would be the brightening of the
Antoine Mérand 2 gravitational microlensing (Einstein, 1936), source.
Françoise Delplancke-Ströbele 2 which was “a little calculation” he had
Andrew Gould 3, 4, 5 carried out 24 years earlier. According to Since 1993, armed with wide-field tele-
Weicheng Zang 6 general relativity, when an object (i.e., a scopes and CCD cameras, time-domain
lens) aligns closely with a background surveys have vastly exceeded Einstein’s
star (i.e., a source) along the line of sight pessimistic expectation and found tens
 avli Institute for Astronomy and Astro-
K to the observer, the light rays from the of thousands of microlensing events so
physics, Peking University, Beijing, source are bent when passing by the lens far by photometrically monitoring almost
China and subsequently form images. In the a billion stars in dense stellar fields (pri-
ESO ideal case of perfect alignment, the bent marily the Galactic bulge). During these
Max Planck Institute for Astronomy, light rays form a ring-like image (called events, the brightness of the source
­Heidelberg, Germany the Einstein ring); more typically, there are ­usually varies over a few weeks as the
Korea Astronomy and Space Science two arc-shaped images with a separation lens star moves with a relative proper
Institute, Daejon, Republic of Korea on the scale of the Einstein ring. For motion of several milliarcseconds per
Department of Astronomy, Ohio State microlensing in the Galaxy, the angular year. Microlensing events have yielded
University, Columbus, USA radius of the Einstein ring (the angular rich astrophysical results, including
Department of Astronomy and Tsinghua Einstein radius) is on the order of only a the discovery of nearly a hundred extra-
Centre for Astrophysics, Tsinghua milliarcsecond. solar planets.
­University, Beijing, China
Einstein’s reluctance to publish was
because he saw little in the way of obser- Microlensing event detected by
Using GRAVITY, we have resolved the vational prospects; he asserted that GRAVITY
two images of a microlensed source “there is no hope of observing this phe-
star for the first time, more than a cen- nomenon directly”. He had two reasons It was not until November 2017 that
tury after Einstein first predicted that for thinking this, both of which relate to microlensed images were successfully
such image splitting could be caused the minuscule angular size of the Einstein resolved for the first time (Dong et al.,
by the gravity of another (lens) star ring. First, for microlensing to occur, 2019), defying Einstein’s dismissal. Our
along the line of sight to the source. two stars need to align within the Einstein team achieved this by observing the
We have measured the angular Einstein ring, and the probability of this is tiny, microlensing event TCP J0507+2447 with
radius (almost exactly half the image no greater than one in a million towards the GRAVITY instrument (GRAVITY
separation) to be 1.87 milliarcseconds, any star in the Galaxy. Second, he antici- ­Collaboration, 2017) on the Very Large
with a precision of just 30 microarc­ pated that no instruments could resolve Telescope Interferometer (VLTI). Our
seconds. The measurement also yields the images, and thus the only observable observations allow us to measure the
the direction of the relative motion of the
lens with respect to the source. These
results, combined with other, so-called Prior: u = 0.2666 ± 0.0036, Flens = 0.2 ± 0.02
microlens parallax measurements, yield
θ E = 1.891 ± 0.014 mas
the lens mass and distance. While this N PA = – 4.9 ± 0.3 deg
lens is an ordinary luminous star, the Flens/Fsource = 0.20 ± 0.02
PA = – 4.9°

same technique could be applied in the u = 0.268 ± 0.03
future to measure the mass and dis- A + = 2.42
χ r2 = 1.20
tance of completely dark objects, such
a black hole. In fact, while black holes 1 E
Δ Declination (mas)

in binaries have been found from X-ray

and LIGO gravitational-wave observa-
tions, and are likely to be found in the

0 S
future by Gaia astrometry, gravitational uθ E = 0.51 mas
microlensing is the only known way to ξ = 0.20
find isolated black holes. Our detection E =

1. Figure 1. Model of the apparent

using GRAVITY on the VLTI opens the –1 89 image. The two red dots are the major
path to such measurements of isolated m and minor images — the sizes of
the dots do not represent the actual
black hole masses. apparent sizes of the images, but

–2 rather an indication of their respective

A – = 1.42
fluxes. The “x” symbol is where the
Introduction unlensed source would be (labelled
“S”). The blue dot is the lens position
with its flux, and the blue dashed circle
In 1936, after persistent prodding by the 2 1 0 –1 –2 is its Einstein ring. Flux is expressed in
amateur scientist Rudi Mandl, Albert Δ Right ascension (mas) fractions of unlensed source flux.

The Messenger 178 – Quarter 4 | 2019 45

GRAVITY Science Dong S. et al., When the Stars Align — the First Resolved Microlensed Images

angular Einstein radius at 2% precision: neous light curves from a space tele- i­nitial discovery. And thanks to the excep-
1.87 ± 0.03 milliarcseconds (see Figure 1). scope in heliocentric orbit. The angular tional site conditions, we were able to
Interferometric resolution of images can Einstein radius measured from the observe it near the magnitude limit of
unleash microlensing’s unique potential VLTI resolution of microlensed images is VLTI/GRAVITY at a relatively high airmass
to find isolated stellar-mass black holes the other missing ingredient that can yield of ~ 1.5.
(BHs) lurking in the Galaxy by lifting the unambiguous determination of the lens
degeneracy between mass and distance mass and thereby definitively identify a The exceptional sensitivity of VLTI/
in the analysis of microlensing light BH lens. GRAVITY and the advent of all-sky bright
curves. transient surveys such as ASAS-SN
The lens of TCP J0507+2447 is not a BH and Gaia provide an unprecedented
LIGO/VIRGO’s astonishing discoveries of but a low-mass star. Nevertheless, it opportunity to obtain more resolved
merging BHs (Abbott et al., 2016) have can serve as a testbed for the above- microlensing images. We hope to carry
raised an important open question: how mentioned approach of lens mass deter- out a systematic survey towards the first
to form BHs with a few tens of solar mination. In an independent effort, definitive identification of an isolated stellar-
masses? Whether they are the end points another research team has measured the mass black hole.
of massive stars or have exotic origins lens flux with Keck adaptive optics
in the early universe, theories predict that images, and by combining this with our
isolated (single) BHs must exist. The precise VLTI angular Einstein radius References
­relative frequency between the single and measurement, they find that the lens is a Abbott, B. et al. 2016, Phys. Rev. Lett., 116, 1102
binary BH populations can provide dwarf star of 0.58 ± 0.03 M⊙ (Fukui et al., Einstein, A. 1936, Science, 84, 506
­crucial clues to the formation mechanism. 2019). Our team (Zang et al., 2019) has Dong, S. et al. 2019, ApJ, 871, 70
However, limited by the detection tech- measured the microlens parallax using Delplancke, F. et al. 2001, A&A, 375, 701
Fukui, A. et al. 2019, AJ, 158, 206
niques, all known stellar-mass BHs are the Spitzer light curves, and by combin- Gould, A. 2000, ApJ, 535, 928
found in binaries. ing the VLTI angular Einstein radius, the GRAVITY Collaboration 2017, A&A, 602, 94
lens is found to be 0.50 ± 0.06 M⊙ . The Nucita, A. et al. 2018, MNRAS, 476, 2962
Microlensing holds great promise in prob- good agreement between the results of Zang, W. et al. 2019, submitted to ApJ,
ing the important yet uncharted parame- these two approaches demonstrates the
ter space of isolated BHs. Estimates robustness of our method. Remarkably,
by Gould (2000) suggest that, amongst around the peak of the light curve of
the microlensing events detected to date, TCP J0507+2447, there was a short-lived
many hundreds may involve BH lenses. anomaly lasting a few hours, suggesting
But thus far only a few BH candidates that the lens star has a 20-Earth-mass
have been reported. This is due to the planet at around 1 astronomical unit
mass-distance-velocity parameter degen- (Nucita et al., 2018; Fukui et al., 2019).
eracy, which makes it impossible to defin-
itively distinguish BHs from low-mass The possibility of using the VLTI to
stars. All existing BH candidates have rel- resolve microlensing images was first
atively long event timescales, which proposed by Delplancke et al. (2001), but
can be due to the large Einstein radii of it had proven to be extremely challenging,
BH lenses with high masses. But a large with numerous failed attempts prior
­Einstein radius can also be produced to our observations. The major challenge
by a low-mass stellar lens at a close dis- had been the difficulty of identifying a
tance. Alternatively, a slow relative proper sufficiently bright target for the interfero-
motion between the lens and source metric observations. A confluence of
stars may induce a long timescale even lucky circumstances facilitated our suc-
with a moderate Einstein radius. cess. Unlike the vast majority of micro-
lensing events found by professional
To completely break the degeneracy, two wide-field surveys towards the Galactic
additional observables are required bulge, TCP J0507+2447 was serendipi-
besides the microlensing event timescale. tously discovered by the Japanese ama-
One is called the “microlensing parallax”, teur astronomer T. Kojima, and the
which depends on the Einstein radius source is at 800 pc towards the Galactic
projected onto the observer’s plane. It anti-centre, making it one of the closest
can be constrained for long events from microlensing events ever found. Our
the distortion of the light curves induced DDT proposal (2100.D-5031) was quickly
by the acceleration of the Earth while accepted, and an ongoing VLTI run
it orbits the Sun or by comparing the allowed our GRAVITY observations to be
ground-based observations with simulta- conducted within about a week of the

46 The Messenger 178 – Quarter 4 | 2019

GRAVITY Science DOI: 10.18727/0722-6691/5177

Hunting Exoplanets with Single-Mode Optical


GRAVITY Collaboration (see page 20)

The GRAVITY instrument was primarily
conceived for imaging and astrometry
of the Galactic centre. However, its 1000
­sensitivity and astrometric capabilities 40
have also enabled interferometry to Declination (mas) 0
reach a new domain of astrophysics: 20
exoplanetology. In March 2019, the 0 0
GRAVITY collaboration published the
first spectrum and astrometry of an 0
exoplanet obtained by optical interfer- –2
ometry. In this article, we show how – 1000 00
this observation is paving the way to –4 Fringe tracker
even more exciting discoveries — find- 00
ing new planets, and characterising
their atmospheres. – 2000
Science spectrometer

New opportunities, new challenges 2000 1000 0 20

Right ascension (mas) 0
With the 2019 Nobel Prize, jointly awarded 10 – 2 0
fibre @ 70 mas
to Michel Mayor and Didier Queloz, the –2 fibre @ 90 mas
Contrast ratio (5σ/15 minutes)

field of exoplanet research received 00

fibre @ 120 mas
worldwide recognition. It is true that for –4 fibre @ 150 mas
10 – 3 00
the first 20 years, the domain was more fibre @ 200 mas
akin to a giant search for Easter eggs. fibre @ 250 mas
The rarity of each discovery meant it had
a major impact. The field was later signi­ 10 – 4
ficantly boosted by the space-based mis-
sions CoRoT (Convection, Rotation and
planetary Transits), Kepler and now TESS
(Transiting Exoplanet Survey Satellite), 10 – 5
resulting in the discovery of thousands of
exoplanets, and the development of a
large community including many young 0 50 100 150 200 250 300
scientists. The success of transit photom- Angular separation (mas)
etry, accompanied by a steady increase
in the capabilities of stable high-precision search for M dwarfs with Exoearths with Figure 1. Upper panels: SPHERE observations of
AU Mic (data from the InfraRed Dual-band Imager
radial velocity instruments — in which Near-infrared and optical Échelle Spec-
and Spectrograph [IRDIS] and the infrared Integral
ESO has invested significantly; for exam- trographs), for example, it is possible Field Spectrograph [IFS]; Boccaletti et al. 2018).
ple, the High Accuracy Radial velocity to constrain the level of atmospheric Over-plotted on the IFS observation are the GRAV-
Planet Searcher (HARPS) and the Echelle evaporation from the observation of a He I ITY single-mode fibres. The sizes of the circles
­c orrespond to the field of view of GRAVITY. The
­SPectrograph for Rocky Exoplanet and line (Alonso-Floriano et al., 2019). How-
fringe-tracker fibre is situated on the star, while the
Stable Spectroscopic Observations ever, transit spectroscopy is limited to spectrometer’s fibre is positioned at separations
(ESPRESSO) — now allows the analysis probing the upper atmosphere, and is between 70 and 250 mas to the south-east of the
of the mass-separation distribution of inherently constrained by the duration of star. From each position of the fibre, a 5-s dynamic
range is extracted and is plotted on the contrast
planets, revealing gaps such as the hot the transit. In the long term, the most
curve. At 120 mas, a dynamic range of 4 × 10 – 5 is
Neptune desert (Neptune-sized planets promising technique is direct spectros- achieved. At 250 mas, the dynamic range is 4 × 10 – 6
within ~ 1 astronomical unit). copy, where the light of the planet (13.5 magnitudes).
(either reflected light or thermal emission)
The next challenge in the field is the is directly imaged on a spectrograph.
­characterisation of exoplanetary atmos- and a numerical capability to remove
pheres through spectroscopy. Until now, This is where GRAVITY enters the field. A the stellar speckles (to distinguish the
the technique has been dominated by good exoplanet imager must have two planet). On an instrument like the Spectro-
high-resolution spectroscopy of evapo- main characteristics: an instrumental Polarimetric High-contrast Exoplanet
rating atmospheres. With the CARMENES capability to remove the stellar diffraction REsearch (SPHERE) instrument, the for-
instrument (Calar Alto high-Resolution pattern (to decrease the photon noise), mer is done by means of a coronagraph,

The Messenger 178 – Quarter 4 | 2019 47

GRAVITY Science GRAVITY Collaboration, Hunting Exoplanets with Single-Mode Optical Interferometry

while the latter uses angular or spectral

differential imaging techniques (ADI b
and SDI). On GRAVITY, an off-axis single-
mode fibre plays the role of a corona-
graph: placed on the planet, its limited
field of view filters out the stellar light.
The GRAVITY interferometer surpasses
single-­dish instruments in post-detection Fringe e
speckle removal: the angular resolution tracker
of about 3 milliarcseconds (mas) yields an Science spectrometer
unprecedented capability to distinguish
speckles from planetary photons. 500 mas
GRAVITY as a planet hunter
GRAVITY’s high dynamic range at Luhman 16A
Exo-REM T = 1150 K / log(g) = 4.3
­angular separations as small as 100 mas GPI data
F λ (10 –14 Wm – 2 µm –1 )

SPHERE photmetry
is obtained thanks to this exquisite post- 0.02
processing. Figure 1 shows ­GRAVITY
observations of the star AU Mic. The disc
of AU Mic has prominent structures, which
are resolved with SPHERE (Boccaletti,
Thalmann & Lagrange, 2015). GRAVITY 0.00
looks for point-like sources, of size
1.9 2.0 2.1 2.2 2.3 2.4 2.5
smaller than its interferometric resolution Wavelength (µm)
(< 3 mas). Larger objects are not seen in
the coherent flux of the interferometer. Figure 2. Upper panel: SPHERE/IRDIS image of CH4 absorption is detected. This gives
HR8799 acquired with a broadband H-filter (from
clues that help to characterise the atmos-
Wertz et al., 2017). As in Figure 1, we put the fringe-
The disadvantage of the single-mode tracker fibre on the star. The science spectrometer phere. The difficulty in interpreting the
interferometer is its field of view, which is fibre is on HR8799e. The sizes of the circles corre- data lies in the complex physical pro-
given by the diffraction limit of the tele- spond to the GRAVITY field of view. Below: GRAVITY cesses at work. Radiative transfer is used
K-band spectrum of HR8799e at spectral resolution
scope (60 mas in the K-band for the VLT to derive the pressure-temperature
500 (grey points) after 2 hours of integration. The
Unit Telescopes). Therefore, while the dashed curve is the K-band Gemini Planet Imager curves, but clouds at different altitudes,
fringe tracker fibre stays on the star, (GPI) spectrum from Greenbaum et al. (2018), show- with various compositions and possibly
the science fibre (which feeds the spec- ing speckle contamination. (GRAVITY Collaboration also heterogeneous, modify the tempera-
et al., 2019).
trograph) is placed at different positions ture distribution. Chemical disequilibrium
across the disc. This is how GRAVITY also adds complexity, with the necessity
hunts for exoplanets; it dithers the posi- GRAVITY as a way to characterise to add chemical timescales and mixing
tion of the fibre to cover a large area. In exoplanet atmospheres coefficients. In short, models need to
the case of AU Mic, we took advantage be challenged by observations, and
of the fact that we are only looking for a In addition to its dynamic range, GRA­ GRAVITY data is meeting that challenge.
planet along an edge-on disc, so we only VITY’s angular resolution yields i) precise
had to scan one line, thereby minimising astrometry (between 10 and 100 μas) In the near future, following the recent
the required telescope time. and ii) K-band spectra mostly unbiased upgrade of GRAVITY’s high-resolution
by s­ tellar light. Fortunately, such near-­ grism, a resolution of 4000 will be achiev-
In the resulting dataset, which covers only infrared spectra are rich in many molecu- able on exoplanets — a significant
the south-eastern part of the disc, no lar absorption lines: for example, H2O, increase compared to the previous reso-
detection was made. The dynamic range CO, CO2, CH4, N2O. We applied this to lution of 500 (because of limited sensitiv-
achieved by GRAVITY, with 15-minute HR8799e in GRAVITY Collaboration et al. ity). GRAVITY will therefore continue to
exposures, is 11 magnitudes at 120 mas (2019). HR8799e is the innermost object challenge models of exoplanetary atmos-
(5-s). At 250 mas the dynamic range is in a multi-planetary system. The angular pheres, requiring simulations with more
even higher, reaching 13.5 mag­nitudes. separation to its host star is 380 mas, resolution and more complex chemical
This is several magnitudes fainter than and the contrast is close to 11 magni- processes. One exciting prospect, for
what was achieved with aperture masking tudes in the K-band. The young planet example, is the detection of C13 isotopes
(Gauchet et al., 2016), and a completely has an effective temperature of 1150 K, (Mollière & Snellen, 2019). In parallel,
new domain compared to what could be still hot from its formation. The spectra, the recent development of atmospheric
done with ADI and SDI techniques on a shown in the bottom panel of Figure 2, parameter retrieval is an exciting new
single 8-metre tele­scope. show the CO absorption bands — no technique, which performs better than

48 The Messenger 178 – Quarter 4 | 2019

fi­ tting a grid of models. The aim is to Acknowledgements Gauchet, L. et al. 2016, A&A, 595A, 31G
GRAVITY Collaboration et al. 2019, A&A, 623, 11
obtain direct estimates of atomic ratios.
See page 23. Greenbaum, A. Z. et al. 2018, AJ, 155, 226G
One of them, the atmospheric C:O ratio, Mollière, P. & Snellen, I. A. G. 2019, A&A, 622A, 139
is currently believed to be a key tracer Öberg, K. I., Murray-Clay, R. & Bergin, E. A. 2011,
of an exoplanet’s formation history References ApJ, 743L, 16O
Wertz, O. et al. 2017, A&A, 598, A83
(Öberg, Murray-Clay & Bergin, 2011).
Alonso-Floriano, F. J. et al. 2019, A&A, 629A, 110
With GRAVITY, we will soon show that Boccaletti, A., Thalmann, C. & Lagrange, A. M. 2015,
we are able to measure this C:O ratio Nature, 526, 230
(GRAVITY Collaboration et al., in press). Boccaletti, A. et al. 2018, A&A, 614A, 52B
ESO/Digitized Sky Survey 2. Acknowledgement: Davide de Martin

Wide-field image showing the field in the constellation

of Pegasus centred on HR8799.

The Messenger 178 – Quarter 4 | 2019 49

Astronomical News

ESO/M. Zamani
Students and project supervisors at the First ESO
Summer Research Programme pose for a photo
at the Supernova Planetarium and Visitor Centre
(see p. 57).

The ESO contract for the design and production of

the cell for the M5 mirror of the Extremely Large
Telescope was signed with SENER Aerospacial
(Spain) on 29 November 2019. The contract was
signed by the ESO Director General Xavier Bar-
cons and the General Director of SENER Aer-
oespacial José Julián Echevarría.

50 The Messenger 178 – Quarter 4 | 2019

Astronomical News DOI: 10.18727/0722-6691/5178

Light Phenomena Over ESO’s Observatories IV:

Dusk and Dawn

Lars Lindberg Christensen 1 Figure 2. Alpenglow

G.Hüdepohl (
seen from above ESO’s
Petr Horálek 1
Paranal Observatory.
domes are coloured red
ESO during the sunset.

Several interesting atmospheric phe-

nomena take place during dusk and
dawn, associated with the setting
and rising of the Sun and Moon. Here,
the most important of these are dis-
cussed in the context of ESO observing
sites. This is the fourth article in a
series about a range of light phenom-
ena that can be experienced at ESO
observatories. The colour and appearance of the sunset quite exciting, cause of exceptional sun-
are very sensitive to clouds and the aero- sets: an astronomical event. In 1908, an
sol content of the upper atmosphere — object — most likely a 100-metre sized
Sunsets no two sunsets are identical. Exceptional meteoroid — exploded over Tunguska
circumstances can have a big impact too; in Siberia (Gladysheva, 2007), and in
Sunsets mark the daily transition from powerful volcanic eruptions, for example, the following period unusual colours at
light to darkness and can be among can increase the dust content at 15–30 km sunset followed by very bright nights
the most beautiful and evocative light altitude in the atmosphere which can were observed in many parts of the world
phenomena that can be seen in the sky a. ­create particularly magnificent sunsets (Kundt, 2001; Longo, 2007).
Everyone on the planet has experienced (Moreno et al., 1965). In 1883, the volcano
the change in light and its effects on Krakatoa erupted in Indonesia and spec-
their mood when the Sun disappears and tacular sunrises and sunsets were subse- Alpenglow
the night begins. quently seen all over the world, glowing
with unusual colours. There have also When the Sun is just below the horizon,
The colours of sunrise and sunset been strong “volcanic sunsets” in living the colour of the sky on the western hori-
are usually very dynamic in the Atacama memory, including those related to zon takes on a warm yellow or yellow-red
Desert. At the Paranal and La Silla Mount St. Helens in 1980, El Chichón in hue and mountains and buildings to the
­observatories, they are enhanced by Mexico in 1982, and Mt. Pinatubo in the east appear to glow red (Figure 2). This
three additional factors: the high altitude, Philippines in 1991–92. The most signifi- phenomenon is named Alpenglow. A
where the air is thinner and clearer; cant sunsets observed in recent times well-known example is Uluru in Australia,
unobstructed views; and the relative were seen worldwide after the eruption whose red colour becomes intensified
vicinity to the ocean, where high levels of of Mount Kasatochi (Waythomas et al., at twilight. Alpenglow can even be seen
humidity lead to strong scattering effects1 2010). They lasted from the end of August in cities.
(Zieger et al., 2013). The sunsets in par- 2008 until at least January 2009. It may
ticular often paint deep colours over the well be considered an irony of nature
horizon (Figure 1). that vol­canoes, which are so powerfully Crepuscular and anticrepuscular rays
destructive, give rise to such beautiful
light phenomena. Most of you will have witnessed the
rays of light that sometimes shine through
Figure 1. Looking east at dusk at the Paranal
Observatory. The pink-coloured Belt of Venus is At least once within the last hundred gaps in cloud cover and illuminate the
followed by the Earth shadow over the horizon. years or so there has been a different, land below. Especially in rainy weather,
R. Wesson/ESO

The Messenger 178 – Quarter 4 | 2019 51

Astronomical News Christensen L. L., Horálek P., Light Phenomena Over ESO’s Observatories IV

Figure 3. The Sun sets

ESO/R. Wesson

over Paranal Observa-

tory, painting an array of
subtle hues across the
sky. Crepuscular rays —
and shadows from the
clouds — are streaming
from the Sun and
appear to converge at
the antisolar point.

these rays can make for a dramatic through the atmosphere, the slower they In 1596, the ship carrying polar scientist
scene. They are also quite often visible become, and thus the more they are Willem Barents (c. 1550–1597) was
at sunset, when they shine over the tops refracted. The effect is measurable — even caught in the ice on what is now called
of clouds or through gaps within the when the Sun is high in the sky — and the Barents Sea at Novaya Zemlya Island.
clouds (Figure 3). They are called crepus- it can be important for navigators and There were two weeks until sunrise, but
cular rays and have been known by vari- astronomers. The extent of the refraction the tip of the solar disc had already peaked
ous poetic names in different cultures, varies from day to day, so the observed above the horizon (de Veer, 1876). Polar
for example, “Maui’s rope” — based on a sunset time can change from one day to explorer Sir Ernest Henry Shackleton
Maori tale; “Buddha’s rays” in parts of the next by as much as 5 minutes. We (1874–1922) also reported repeated
south-east Asia; and “Jacob’s ladder” cannot therefore rely on an ephemeris to sunrises “ahead of time” in the Arctic
in the UK (Lynch & ­Livingston, 2001). predict the observed sunset time with
perfect accuracy. Figure 4. During the early evening of 7 August 2017,
a partial lunar eclipse was visible in the sky above
When the Sun sets behind high moun- the ESO Headquarters. While the Moon was rising,
tains to the west, the mountains create A more specific variation of this phenom- significant anticrepuscular rays were visible in the
broad crepuscular rays by partially block- enon is visible in the Arctic regions. antisolar direction.
ing the light. Some dust or haze in the
air will serve to highlight the rays. Under

ESO/P. Horálek
perfect conditions, when crepuscular
rays can be followed all the way across
the sky, they appear to expand at first
and then narrow towards the east. In
reality, they remain in parallel but only
appear narrower at greater distances
from us. Crepuscular rays in the east are
called anti-crepuscular rays (Figure 4).

Refraction, differential refraction and

the green flash

There is a whole subset of interesting

phenomena related to the shapes of
the Sun and Moon as they rise and set.

Few people are aware that we still see

the Sun for a few minutes after it has
moved below the horizon. The reason we
still see it in the sky is that its rays are
refracted by the atmosphere, raising the
image of the solar disc by about a dia­
meter on average as it sets 2. In other
words the Sun’s apparent movement
towards the horizon is slowed down by
the refraction.

The refraction happens because light rays

travel more slowly as they pass through
the Earth’s atmosphere. The lower the
Sun is, the further the light rays travel

52 The Messenger 178 – Quarter 4 | 2019

Figure 5. The setting rise to a Chinese lantern effect in which
P. Horálek/ESO

crescent Moon
the solar or lunar disc appears layered.
deformed by the atmos-
phere over the Pacific The subtle details of the layered lantern
Ocean, as seen from effect can best be observed with the
Paranal on 19 November Moon (see Figure 5), as the Sun often
2017. The effect of layers
saturates on pictures. The “lifting” effect
with different tempera-
ture and density in ­ of the lower layers of the atmosphere
the atmosphere caused can also be seen on stars as they set
different parts of the (Figure 6).
Moon’s image to be
refracted by different
amounts as it neared Compared to other natural light phenom-
the horizon. ena, the green flash has an aura of mys-
tery and supernaturality. The green flash
(Shackleton, 1919). The explanation is Another effect of refraction is that the Sun is rare and very difficult to observe, but,
once again the atmosphere’s refraction of and Moon never appear perfectly round when the conditions are right, the observ-
light or rather, the variation in that refrac- as they approach the horizon, they er’s perseverance is rewarded with a
tion. In the case of Barents, the refraction appear flattened — as also evidenced in quick green gleam at the top of the set-
must have lifted the solar disc by about many of the photographs in this article. ting Sun, amid the red and orange shades.
5 degrees to peek over the horizon, The amount by which the rays are
but this is one of the more extreme refracted increases closer to the horizon, At sunset the blue and green light rays
cases. The phenomenon is known as the so rays from the upper edge of the solar are refracted a little more than the red
Novaya Zemlya effect (Lehn, 1979; Lehn & disc are refracted less than those from light, which means that the blue-green
German, 1981). the lower edge, flattening its shape. colours are “lifted” slightly more in the
Occasionally, the shape of the disc may sky. You could almost say there are two
be further disturbed if the layers in the solar discs — one blue-green and one
Figure 6. Several effects are seen on the setting
stars in this multi-exposure photo: atmospheric
atmosphere have different temperatures red (Young, 2013). Where they overlap,
­scattering (reddening) and absorption, and differen- and refract the light by different amounts we see a yellowish­-red Sun, with the red-
tial refraction (“lifting” of the stars). — called differential refraction. It gives dening of the Sun due to atmospheric

R. Wesson/ESO

The Messenger 178 – Quarter 4 | 2019 53

Astronomical News Christensen L. L., Horálek P., Light Phenomena Over ESO’s Observatories IV

scattering. With the right equipment it is

P. Horálek/ESO
not so rare to see the Sun with a greenish
top, called the green segment, even before
it reaches the horizon (Figures 7 and 8).

However, differential refraction alone does

not result in more than the green and
red segments. To produce a proper flash Figure 7 (above). This sequence taken at La Silla on Figure 8 (below). An example of a green segment
16 November 2017 shows the phenomena of red seen from Cerro Paranal. The image was taken by
such as those seen in Figures 8 and 9,
(below the solar disc) and green flashes (above) that Stéphane Guisard (ESO). Light phenomena connois-
the red light is spectrally separated from can occur when the Sun sets and the atmosphere seurs argue that this is strictly speaking not the
the blue-green light. Without spectral refracts the sunlight into different colours. green flash.
separation, only a green (upper) or a red
(lower) fringe would be seen in the solar

ESO/S. Guisard
image. The ozone layer is responsible for
the spectral effect, since the light path
through the atmosphere is longer at
­sunrise and sunset and the ozone layer
covers altitudes between about 12 and
40 km. The ­Chappuis band of ozone
absorbs light in a broad band centred
close to 590 nm in the orange. Owing to
the long path, the effect from this band
strengthens when the Sun is near the
horizon, effectively removing much of the
yellow, orange and red light. In the right
conditions this can produce the apparent
spectral separation needed to produce a
noticeable flash.

Simple atmospheric modelling 3 shows

that in order to see a green flash at all,
the aerosol content of the atmosphere
needs to be very low where the sunlight
grazes the horizon. It also turns out that

G. Blanchard/ESO
in rare circumstances, when the light
path is guided by differential refraction
to take an unusually long track through
the ozone layer, the flash is significantly
shifted towards shorter wavelengths to
produce the magnificent (and very rare,
Figure 9) blue flash; lucky is the person
who witnesses that!

Dutch researcher Marcel Minnaert (1893–

1970) saw the green flash shoot up like
a flame from the horizon (Minnaert, 1993).
Because the green-blue rim on the solar
disc is very thin (just a few arc-seconds),
it is only alone over the horizon for approx-
imately one second. Accordingly, the
green flash is seen to last for a similarly
short length of time at ESO observatories.
It can, however, last much longer. Polar
explorer Richard E. Byrd (1888–1957) and
his crew claimed to have seen a “green
sun” persist for up to 35 minutes while on
an expedition to Antarctica in 1929 (Lock,
2015). Despite this unique observation, Figure 9. ESO staff member Guillaume Blanchard observing the sunset on Christmas Eve 2007 from
there are relatively few pictures of the was able to capture the very rare blue flash while the Paranal Residencia.

54 The Messenger 178 – Quarter 4 | 2019

phenomenon, which contributes to the Figure 11. A glory

ESO/C. Malin
caused by sunlight
whole mystery surrounding the flash
backscattering off tiny
and makes it something quite special to droplets of water in
look for. the atmosphere. Glories
appear at a point
directly opposite the
The green flash is best seen on a com-
position of the Sun,
pletely unobstructed western horizon, so they are only visible
like the view from the Paranal platform. at sunrise or sunset.
The weather must be very clear, and
the atmosphere needs to have complex
layers and a low aerosol content.
Happy hunting!
the horizon upwards as the Sun goes atmosphere that lie on the horizon or
down. The twilight arch is the arc of light high above our heads. The Earth shadow
The Earth shadow and the Belt of Venus that forms over the place where the Sun emerges and its visibility is best when
has set. It is usually red at the bottom, there is little dust or haze in the air.
As the Sun reaches the horizon, an yellow for a wide stretch above the red,
orange-yellow twilight arch forms to the and arches over in a peach-coloured, The clarity and low humidity of the air at
west, and the blue-grey Earth shadow green, turquoise or slightly purple band the high altitudes of the Atacama Desert
slowly rises to the east; stretching from that merges with the background colour. provide extraordinary opportunities to
It is created by scattered sunlight in the regularly observe the phenomenon called
Figure 10. The pink Belt of Venus and Earth shadow atmosphere. Even when the Sun has set, the Belt of Venus, followed by the projec-
as seen at ALMA. it can continue shining on parts of the tion of the dark Earth shadow onto the
D. Kordan/ESO

The Messenger 178 – Quarter 4 | 2019 55

Astronomical News Christensen L. L., Horálek P., Light Phenomena Over ESO’s Observatories IV

Figure 12. Looking west just after sunset at ESO’s

ESO/P. Horálek
Paranal Observatory on 25 January 2015. The bright
object is Venus. In this view rich dusk colours can
be seen. They were likely caused by volcanic ash
from the January 2015 eruption of Tongan volcano
and possibly even the 2014 eruptions of the Indone-
sian volcano Mount Sinabung.

atmosphere (Figure 10). Looking towards

the antisolar point some minutes after
the sunset or before the sunrise, the sky
over the horizon seems like a dark curtain
bounded by pink, while the dusk or dawn
sky above is much brighter.

Seen from the ground when the Sun is

below the horizon, the Belt of Venus is
the result of light from the setting or rising
sun being backscattered by fine dust
particles and aerosols that are present
higher in the atmosphere (Lee, 2015).
It is most easily visible just a few minutes
after sunset or before sunrise. The belt
appears as a glowing, pinkish arch that
extends roughly 10–20 degrees above
the horizon.

The glory
est stars in the sky, can be visible in the 3
The telluric spectrum of the green flash (Fosbury, R.
Observers at La Silla or Paranal can at sky very early after sunset (Figure 12) or
times see a phenomenon called a glory very shortly before sunrise. Usually just 4
Green and red rims (Young, A. T. 2013): https://aty.
when looking down on a cloud layer. minutes after sunset the brightest stars
Glories are concentrated coloured rings appear while the Belt of Venus becomes
around the shadow of the observer’s larger and the sky above is “swallowed”
head (or the shadow of a camera; see by the Earth’s shadow. References
Figure 11). They occur by backscattering
Christensen, L. L. et al. 2016, The Messenger, 163, 40
on tiny water droplets in clouds or fog at Gladysheva, O. 2007, Solar System Research, 41, 314
the antisolar point and look like circular Acknowledgements Horálek, P. et al. 2016a, The Messenger, 163, 43
rainbows (Nussenzveig, 2011). This is a Horálek, P. et al. 2016b, The Messenger, 164, 45
The authors are grateful for helpful conversations Kundt, W. 2001, Current Science, 81, No. 4, 399
rather complicated case of Mie scattering with Bob Fosbury. Sarah Leach and Laura Hiscott Lee, R. L. 2015, Applied Optics, Vol. 54, Issue 4, B194
(and not the special divine importance are thanked for improvements to an earlier version Lehn, W. H. 1979, Journal of the Optical Society of
of a person!). If two people are standing of the text. America, Vol. 69, Issue 5, 776
on a mountain and look at both of their Lehn, W. H. & German, B. A. 1981, Applied Optics,
Vol. 20, No. 12, 2043
shadows, they will each see only one
Lock, J. A. 2015, Applied Optics, Vol. 54, Issue 4, B54
glory and claim it to be around their own Notes
Longo, G. 2007, Comet/Asteroid Impact and Human
head. Looking down at a plane’s shadow a
Society, 303
 lease don’t forget that looking at the Sun itself,
when flying, you will often be able to see especially through an optical device (camera, tele-
Lynch, L. K. & Livingston, W. 2001, Color and Light in
Nature, (Cambridge: Cambridge University Press)
a glory on the cloud tops. scope, binoculars, etc.), is very dangerous, and Minnaert, M. G. J. 1993, Light and Color in the
could cause immediate blindness. Do not attempt Outdoor, (New York: Springer)
to observe the Sun unless you know what you are Moreno, H. et al. 1965, Science, 148, 364
Planets in dusk and dawn Nussenzveig, H. M. 2012, Scientific American, 306, 68
Shackleton, E. H. 1919, South: The Story of
Shackleton’s 1914–1917 Expedition, (London:
The clear air at the high altitudes of the Links ­W illiam Heinemann)
observatories makes dawn and twilight 1
de Veer, G. 1876, The Three Voyages of William
The colors of sunset and twilight (Corfidi, S. F.
colours very intense, but also followed by 2014, NOAA/NWS SPC): https://www.spc.noaa.
Barents to the Arctic Regions: 1594, 1595 and
1596, (London: Forgotten Books, 2017)
a steep darkening gradient. For these gov/publications/corfidi/sunset/ Waythomas, C. W. et al. 2010, Journal of
reasons, very bright objects, such as 2
Effect of atmospheric refraction on the times of Geophysical Research, 115, B12
planets in our Solar System or the bright- sunrise and sunset (Tong, Y. 2017, HKO): https:// Zieger, P. et al. 2013, Atmospheric Chemistry and Physics, 13, 10609

56 The Messenger 178 – Quarter 4 | 2019

Astronomical News DOI: 10.18727/0722-6691/5179

The ESO Summer Research Programme 2019

Carlo F. Manara 1 100 Figure 1. Distribution of

the nationalities of the
Christopher Harrison 1
applicants for the first

Number of applicants
Anita Zanella 1 ESO Summer Research
Claudia Agliozzo1 Programme.
Richard I. Anderson 1
Fabrizio Arrigoni Battaia 2
Francesco Belfiore 1 50
Remco van der Burg 1
Chian-Chou Chen (T. C.) 1
Stefano Facchini 1
Jérémy Fensch 1
Prashin Jethwa 1 0
Rosita Kokotanekova 1










Federico Lelli 1
Anna Miotello 1
Anna Pala 1
Miguel Querejeta 1
Adam Rubin 1 Nationality
Dominika Wylezalek 1
Laura Watkins 1 as an opportunity that had been missing Programme overview
at Garching until now and decided to
organise a six-week-long Summer The programme started with a workshop,
ESO Research Programme at ESO for up to open to all ESO staff, on 1 July 2019.
Max Planck Institute for Astrophysics, seven university students. At this workshop the seven research pro-
Garching, Germany jects were introduced by the advisors,
A proposal was submitted by the and the students introduced themselves.
Garching ESO Fellows requesting funds An introduction to ESO was delivered
For the first time ever, a summer from the Director for Science to cover by the head of the ESO User Support
research programme was organised at travel costs and to provide a basic stipend Department Marina Rejkuba, and the
ESO Garching. Seven students, enrolled to cover lodging and living expenses for Director General greeted all participants
in universities all around the world, the students. The proposal was accepted from the control room of the La Silla
were selected from more than 300 and ESO Fellows, with the support of Observatory (he was visiting La Silla for
applicants. They each spent six weeks ESO administrative assistants, organised the total solar eclipse at the time).
from June to August 2019 carrying out the first-ever ESO Summer Research
a scientific project under the supervi- Programme. This involved booking ESO The students were each working on their
sion of teams of ESO Fellows and post- apartments and office space, setting up own research project, with the supervi-
docs, while engaging in the scientific the website 1, organising the application sion of one or more ESO Fellows, for the
life of ESO. The students carried out process and the selection of students, duration of the programme (Figures 2 & 3).
research in different fields of astron- planning and delivering a lecture series The schedule in the first three weeks also
omy, from comets to high-redshift gal- and, most importantly, designing and consisted of a set of eight lectures on
axies and from pulsating stars to leading the research projects. astronomical topics, a visit to the ESO
protoplanetary discs. In this report we Supernova including a planetarium show,
present the programme and describe The response from the community was and a telecon with Anita Zanella (an ESO
the main outcomes of the projects. incredible. More than 300 valid applica- Fellow observing at Paranal). The final
tions were received from university stu- three weeks were mainly focused on
dents in most Member States, from our the research projects, but with an addi­-
Motivation and organisation Host Country Chile, and from ESO’s stra- tional two lectures and one visit to the
tegic partner, Australia (Figure 1). Partici- Extremely Large Telescope (ELT) primary
Summer studentship programmes for pants were selected by first distributing mirror test stand. Most of these additional
undergraduate students are becoming the applications amongst all potential activities and lectures were organised
the preferred way for an enterprising stu- supervisors for an initial ranking, followed ­following an explicit request from the stu-
dent to gain their first research experi- by a final selection by a committee com- dents who had expressed enthusiasm
ence; these programmes can last from a prising three fellows, one student and about the first set of lectures. Throughout
few weeks to months at top-class inter- one staff member. The final list included the duration of the programme the stu-
national universities or research centres. seven students — four females and three dents were among the most active
Such programmes have a wide range of males — from seven different countries. attendees of scientific activities at ESO
benefits to students and hosts alike. The After a short video interview all seven stu- Headquarters, including talks, science
ESO Fellows in Garching identified this dents accepted the offer. coffees, and informal meetings.

The Messenger 178 – Quarter 4 | 2019 57

Astronomical News Manara C. F. et al., The ESO Summer Research Programme 2019

Figure 2. Student Tania Machado with her Figure 3. ESO Summer Research Programme stu-
­supervisor Chris Harrison. dent Aisha Bachmann with supervisors Jeremy
Fensch and Remco van der Burg.

The last days were all focused on the This is a very challenging task because them during the next months and write
preparation of the most thrilling event for only the Hubble Space Telescope (HST) up her findings in a publication.
the students: their 15-minute presentation can spatially resolve UDGs at high red-
to be given in front of ESO staff, students shift, and cosmological surface bright-
and fellows during the final workshop ness dimming makes them extremely dif- Comet evolution from the Kuiper Belt
in the old ESO auditorium. This event ficult to detect. to a dormant comet in the near-Earth
was very well attended by ESO personnel asteroid population
(Figure 4) and showcased the great Aisha looked for UDGs, at redshifts Advisors: Rosita Kokotanekova
­science that the students were able to beyond 1, in the deepest cluster images Student: Abbie Donaldson (UK & Ireland),
achieve during this relatively short pro- that were ever taken with the HST. She University of St Andrews, UK
gramme; some examples are described wrote a detection algorithm and tested it
in the next section. on mock galaxies that she inserted into This project focused on analysing photo-
the data; she then used the algorithm to metric observations of the comet 169P/
search for real UDG candidates. Finally, NEAT taken between February and June
Students and their research projects Aisha identified which UDGs, among the 2019 with the FOcal Reducer/low disper-
candidates she found, are cluster mem- sion Spectrograph 2 (FORS2) on the Very
Understanding the formation mechanism bers rather than projections along the line Large Telescope (VLT) and with the Wide
of galaxies at their extremes of sight by statistically comparing her Field Camera (WFC) on the Isaac Newton
Advisors: R
 emco van der Burg & Jérémy detections with those of a reference field. Telescope (INT) on La Palma. Since the
Fensch Her preliminary results look extremely comet was observed close to aphelion
Student: Aisha Bachmann (German), interesting and Aisha aims to finalise and was therefore inactive, the photomet-
­University Bochum, Germany

One of the most surprising recent results

in the field of galaxy formation is the
­discovery of a significant population of
ultra-diffuse galaxies (UDGs) in local gal-
axy clusters. These are galaxies of the
size of the Milky Way, but with a stellar
mass similar to dwarf galaxies. Theorists
are proposing models that can produce
such galaxies in simulations; these gen-
erally invoke tidal heating scenarios
­arising from interactions with neighbour-
ing galaxies, or outflows coming from Figure 4. One of the
the galaxies themselves. To distinguish research students,
­Matthew Wilkinson,
amongst these different scenarios it is ­p resents his research to
important to study the abundance of fellow participants and
UDGs as a function of cosmic epoch. ESO staff.

58 The Messenger 178 – Quarter 4 | 2019

169P/NEAT, P = 8.381 hours Figure 5. Rotational work on the project by investigating
15.4 light curve of comet
­similar data sets in other ELAN fields, as
169P/NEAT derived
from INT/WFC data. well as enjoying a trip to Hawai’i to carry
15.6 The different symbols out observing runs at the James Clerk
correspond to data Maxwell Telescope (JCMT) on Maunakea.
taken during each of the
H r (mag)

15.8 six observing epochs

between February and
May 2019. Modulated variability: a new window
16.0 into stellar pulsations
Advisors: Richard I. Anderson
Student: Samuel Ward (UK), University of
Durham, UK
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Rotational phase What causes the variability patterns of
classical Cepheid variable stars to
ric observations could be used to extract ing standard analysis techniques and not change over time? More and more modu-
the brightness variation of the nucleus ­limited by the data quality. Work is lated variability is being discovered
due to rotation and change of geometry. now required to optimise the techniques among Cepheids, yet its origin remains
Abbie derived the rotational light curve before the arrival of the exquisite elusive, and the properties of the modula-
of 169P/NEAT using the most likely rota- ­HARMONI data. Tania has a strong inter- tion challenge the classical paradigm
tion period of 8.381 hours (Figure 5) est in keeping in contact with her advi- of Cepheids as other well-understood,
and constrained the comet’s albedo and sors and with ESO to continue this work more simple, variable stars.
the slope of the phase function. Abbie’s and she hopes to show her results
results will be included in a publication at the conference “Spectroscopy with Samuel analysed an 8-year-long set of
comparing the surface properties of two ­HARMONI at the ELT” to be held in high-resolution optical spectra of a bright
of the darkest Jupiter-family comets, Oxford in September 2020. Cepheid to unravel the nature and cause
169P/NEAT and 162P/Siding Spring with of the star’s modulated variability. He cre-
other comets and asteroids. ated his own method for modeling spec-
Caught in the act: witnessing the forma- tral line profiles using multiple compo-
tion of the most massive galaxy clusters nents and used it to trace the changes in
Preparing for the Extremely Large Tele- across the cosmic time complex line profiles over time. Addition-
scope: how will high-redshift star-form- Advisors: C  hian-Chou Chen (T. C.) & ally, he investigated how different atmos-
ing galaxies appear with HARMONI? ­Fabrizio Arrigoni Battaia, pheric layers move at different velocities.
Advisors: Anita Zanella & Chris Harrison Student: Marta Nowotka (Polish), Samuel found compelling evidence that
Student: Tania Machado (Portugese), ­Colorado College, USA the observed modulated line splitting
Technico Lisbon, Portugal is most likely caused by non-radial pulsa-
In the hierarchical model of structure for- tion modes, rather than by atmospheric
The ELT, with its 39-metre diameter pri- mation, the most massive galaxies often shock related to the dominant pulsation
mary mirror, will have the angular resolu- form through merging processes within mode, as previously proposed.
tion and light gathering power to revolu- the highest density peaks, known as pro-
tionise our understanding in many toclusters. Identifying these protoclusters
astrophysical fields. This project is in and characterising their properties is key Dark matter content of galaxies from
preparation for the use of the High Angu- to reaching a full understanding of galaxy globular cluster kinematics
lar Resolution Monolithic Optical and formation. Recently, new prime candi- Advisors: P
 rashin Jethwa & Laura
Near-infrared Integral-field spectrograph dates for signposts of massive protoclus- Watkins
(HARMONI), a first-generation ELT ters have been discovered: they are enor- Student: Matthew Wilkinson (Australian),
­integral-field spectrograph, to spatially mous (> 200 kpc) Lyman-a nebulae University of Queensland,
resolve the interstellar medium of high- (ELAN) which host multiple active galactic Australia
redshift (z ~ 2–5) galaxies and to meas- nuclei and are surrounded by over-densi-
ure the physical processes occurring on ties of Lyman-a emitters. How well can we measure the amount
scales of individual star-forming regions. of dark matter in a galaxy? This was the
Tania created simulated HARMONI data- To better understand their formation central question of this project, and its
cubes of how galaxies at z ~ 2–3 will ­history, Marta developed complex ­ answer will have important consequences
appear when observed with different Python algorithms to analyze SCUBA-2 for our understanding of cosmology and
observing strategies and observing con- 850-micron data and found evidence galaxy formation. Questions about dark
ditions. The most important result of dust-obscured star formation around matter certainly motivated our pool of
from Tania’s work was that our ability to one ELAN. This is an exciting result and potential students, with 115 eager appli-
extract key physical properties from we expect it to be published in high-­ cants for this project. Out of this t­ alented
the simulated data was limited by apply- impact journals. Marta will continue to pool, we selected Matthew, who tested

The Messenger 178 – Quarter 4 | 2019 59

Astronomical News Manara C. F. et al., The ESO Summer Research Programme 2019

the accuracy of calculations of galactic Response (5 max) Figure 6. Average

results from the feed-
dark matter content. 0 1 2 3 4 5
back given by students
Enjoyed the programme at the end of the pro-
Matthew tested calculations which use Likely to recommend gramme; 5 is the most
observations of globular clusters — positive score.
dense, bright clusters containing tens of Useful lectures
thousands of stars. This very same calcu- Useful research project
lation had recently been applied to obser- Interaction with other ESO staff
vations in the Milky Way, so testing its
accuracy is of real, present importance. Useful programme overall
To do this test, Matthew applied the cal- Knowledge of what ESO does
culation to simulations and compared the
Organisation of programme
results to the correct answer known from
the simulations. The results suggest that
the calculation may be underestimating and Upper Sco, two star-forming regions ment and we are pleased to report that
the amount of dark matter in galaxies. spanning ages between 2 and 10 Myr. funding has been secured to run the pro-
This is a tentative result and confirming it Francesco developed a code aimed at gramme again in the summer of 2020.
would require more tests. Prashin and reproducing the observed CO fluxes This is great news for many, including
Laura remain in touch with Matthew and within the viscous evolution framework, future potential applicants who have
are enthusiastically supporting him as he with interesting results. While the model already started to inquire about the dead-
applies for PhD positions in astronomy. reproduces the statistical properties of line for applications. How this programme
individual star forming regions well, it continues will depend on the efforts of
is not able to fit all of the star forming many, and its expansion to include more
Testing disc evolution with ALMA surveys regions simultaneously. This suggests ESO staff, including Fellows in Santiago
of CO emission either that the viscous evolution scenario and/or other ESO departments, is very
Advisors: S tefano Facchini, Anna ­Miotello has to be revisited, or that the two star much encouraged.
& Carlo Manara forming regions had different initial condi-
Student: Francesco Zagaria (Italian), tions in their disc mass and radius distri-
­University of Pavia, Italy butions. The results are presented in a Acknowledgements
draft paper that will be submitted soon. The ESO Fellows and postdocs in Garching
How protoplanetary discs evolve is a acknowledge the active support and encouragement
long-standing question. How they evolve of the Director for Science Rob Ivison, the ESO Fac-
determines the planet formation potential Feedback and future programmes ulty, and several ESO students, staff and administra-
tive assistants. In particular, we thank Stella-Maria
of discs and is a key ingredient in any Chasiotis-Klingner for her excellent support w ­ ith
planet formation model. The two main We asked students to give feedback on logistics. We acknowledge funding from the Directo-
theoretical paradigms describe disc evo- the programme, and the responses were rate for Science to cover the travel costs and sti-
lution as driven by viscosity, or by mag- extremely positive (Figure 6). Interviews pends for the students.
netically supported winds. The two lead carried out with the students are pre-
to different predictions about the evolu- sented in the ESOBlog 2 and highlight Links
tion of gas disc radii, with the former pre- how much they enjoyed their research
dicting that the disc radii should expand. experience and the programme overall.  SO summer research programme website:
In this project, Francesco tested the vis- 2
ESOBlog entry Meet Our 2019 Summer Research
cous scenarios by comparing statistical The great success of the programme Programme Students:
properties of CO fluxes measured by has not been overlooked by the ESO blog/from-comets-to-cosmology/
ALMA for the disc populations of Lupus Director of Science and by ESO manage-
ESO/S. Otalora

The ALMA array on the

Chajnantor Plateau from
October 2019.

60 The Messenger 178 – Quarter 4 | 2019

Astronomical News DOI: 10.18727/0722-6691/5180

Report on the ESO Workshop

Artificial Intelligence in Astronomy

held at ESO Headquarters, Garching, Germany, 22–26 July 2019

Henri M. J. Boffin 1 As AI methods become more commonly

Tereza Jerabkova 1, 2, 3 used, a fundamental understanding of
Antoine Mérand 1 their limitations, assumptions, and perfor-
Felix Stoehr 1 mance is due. The rigour of the scientific
process requires that such methods
are applied with extreme care and to
ESO ensure that “the machine is being taught
Helmholtz-Institut für Strahlen- und to take into account certain relevant
Kernphysik, Universität Bonn, Germany known facts” (Griffin 2014). Moreover, the
Astronomical Institute, Charles perspectives of information theory,
­University, Prague, Czech Republic neural science, and other areas on AI are
expected to stimulate and guide the
development of the next generation of
In July 2019, ESO hosted one of the first intelligent methods used in astronomy
international workshops on artificial and elsewhere. It was therefore thought
intelligence in astronomy, with the dou- to be an ideal time to host an interna-
ble aims of presenting the current tional workshop on AI in astronomy. ESO
landscape of methods and applications organising such a workshop was impor-
in astronomy and preparing the next tant for several reasons, but perhaps
generations of astronomers to embark less known is that ESO has had an inter-
on these fields. In addition to a wide est in AI for a long time (for example,
range of review and contributed talks, Adorf, 1991).
as well as posters, the ~ 150 attendees
could learn the techniques through sev- Artificial intelligence covers a wide range Figure 1. Conference poster.
eral dedicated tutorials. of algorithms and methods. The first goal
of the workshop was to provide a clear
map by which to navigate this jungle and ­ lustering and dimensionality reduction.
It is certainly an understatement that arti- show which techniques are used for These can be done through a variety
ficial intelligence (AI), i.e., intelligence which kind of science. This was done in of methods and it is important to know
demonstrated by machines, has taken a few invited talks by prominent speakers which one is best suited to a given prob-
the world by storm, with breakthroughs to clearly set the scene. Laura Leal-Taixé lem. He presented several applications,
appearing in the news almost daily. provided an introduction to deep learning such as how to find 101 new stellar clus-
Indeed, the incredible progress in com- using computer vision and its application ters in the Milky Way or several lensed
puter power, the availability of large to autonomous driving as a clear example quasars thanks to Gaia or how to dis-
amounts of data and the ability to pro- that will affect our lives more and more. cover hundred of thousands of new gal-
cess them (even if they are unstructured), She stressed the importance of increas- axies by combining WISE and Gaia data.
coupled to a theoretical understanding ing the diversity in the data, but also in These results would have been hard, if
of techniques such as machine learn- the community which is building the not impossible, to obtain without the use
ing, and, more generally, data mining, ­algorithms, in order to avoid biases (that, of machine learning, but it is important
have allowed AI to advance at a frantic in daily life, can be fatal). Emille Ishida to note that while unsupervised learning
rate, including in science. Astronomy showed the need for active learning in is powerful, it also requires some taming
is no exception. The sheer volume of astronomy, as the basic assumption of and a clear understanding of the limita-
astronomical data (which is increasing supervised machine learning — that the tions. That this is true was reviewed by
exponentially; see for example, Stoehr, training set is fully representative of the Giuseppe Longo, who also explained
2019) necessitates a new paradigm. Data target sample — is often not fulfilled. It is that the use of these methods will lead to
analysis must become, to a large extent, thus essential that humans intervene in often unexpected results that make
more automated and more efficient, in the process to enhance the efficiency sense to humans only a posteriori. Dalya
particular through AI. And this is indeed of the learning. She stressed the need for Baron showed in impressive fashion how
what is happening. A look at the NASA optimised samples and algorithms for one can mine for novel information in
Astrophysical Data System shows that machine learning applications and that large and complex datasets using outlier
before 2005 only 21 refereed papers had interdisciplinarity is the key, but also that detection and dimensionality reduction
“machine learning a” in their abstract. serendipitous discoveries will only get algorithms (Baron, 2019). Particularly
Since then the number has been multi- more difficult with the next generation impressive was how such techniques
plied by 41, with 663 papers published of large-scale surveys and it is therefore allowed astronomers to find a new corre-
within the last five years, in an almost essential to plan for the unknown with lation amongst active galactic nuclei.
exponential way (there were twice as adaptable algorithms. Alberto Krone-
many published in 2018 than in 2017, for Martins presented the other way to Finding the right method is not an
example). achieve results in astronomy: with unsu- easy task and it is important to bring
pervised learning, that is, essentially together experts from different fields.

The Messenger 178 – Quarter 4 | 2019 61

Astronomical News Boffin H. M. J. et al., Report on the ESO Workshop “Artificial Intelligence in Astronomy”

Figure 2. Conference photo. Zdenka Kuncic presented a special, and As already indicated, there were also
quite inspiring, talk about emergent four three-hour tutorials and hands-on
Mi Dai ­presented the Photometric LSST ­intelligence from neuromorphic complex- sessions that allowed the participants
Astronomical Time series Classification ity and synthetic synapses in nanowire to delve directly into the techniques. These
Challenge (PLAsTiCC) and described networks. After presenting a brief history covered an introduction to machine learn-
how to involve the community at large of AI, she showed that to reach the ing using Python notebooks, machine
and its thousands of machine learning ­ultimate goal of general intelligence, one learning and deep learning using disitrib-
experts via, for example, the Kaggle needs to move away from mainstream uted frameworks and optimised libraries,
­platform 2, to come up with the best algo- computing. She told us that companies and how to use NIFTy.
rithms to classify the very many transient are already developing sophisticated
sources that will be found by the Large neuromorphic chips, which consume The workshop closed with a final discus-
Synoptic Survey Telescope. Similarly, orders of magnitude less power than sion led by Torsten Enßlin which proved
Rafael de Souza presented the Cosmos- conventional processors and which try to that AI is needed in astronomy and will be
tatistics Initiative , an endeavour aimed at emulate the brain. She also described even more in the future, especially as we
fostering inter­disciplinary collaborations how scientists and engineers are now won’t be able to store all the data and
around astronomy and characterised by creating biomimetic structures of nano­ on-the-fly decisions will have to be made.
a residence programme. wires that self-assemble into a complex, It was also stressed that interdisciplinary
densely interconnected network, with a teams are required, as well as a new kind
It was also important to make sure that topology similar to a biological neural of physicist who will have to be trained.
the ground covered by the workshop was network and characterised by a collective The need to better understand the meth-
as wide as possible. Accordingly, John memory. ods that are used was also stressed —
Skilling spoke on how to do computation as scientists, we shouldn’t rely on “black
in big spaces, presenting the framework This very impressive series of invited talks boxes” and need to be very critical.
of inference, i.e. the Bayes therorem, and was complemented by numerous con- This requires us to learn the language of
how the prior space is often much bigger tributed talks and posters, covering the the data scientists and the basic under-
than the small posterior space, leading whole range of applications of AI meth- pinning of the methods, such as Bayesian
to a lot of confusion. Only by reducing ods in astronomy, from meteorite hunting probability.
dimensionality can one hope to solve the to augmenting N-body simulations with
problems. Jens Jasche showed how to deep learning models, through applica- The workshop was a great success and
perform large-scale Bayesian analyses tions in adaptive optics. A poster compe- participants praised the overall quality
of cosmological datasets, using computer tition was organised, and participants of the talks and tutorials, as well as of the
programs and not analytic functions to were asked to vote for the best posters. abstract booklet 4. Many were already
perform a hierarchical Bayes analysis. The three winners each received a hoping that a related workshop would
This allows one, for example, to infer the mounted ESO image; they were: Philipp take place next year! We therefore invite
mass density in a super-galactic plane or Baumeister (Using Mixture Density Net- the community to organise such events
estimate galactic cluster masses. In the works to Infer the Interior Structure as regularly as possible. The PDFs of all
same vein, Torsten Enßlin presented both of Exoplanets); Timothy Gebhard (Learn- the talks and posters and the material of
in a contributed talk and in a tutorial the ing Causal Pixel-Wise Noise Models two of the three tutorials are avail­able on
fully Bayesian information field theory and to Search for Exoplanets in Direct Imag- the workshop webpage 1. All in all, the
the Numerical Information Field Theory ing Data); and Colin Jacobs (Using Deep talks, tutorials and posters covered a very
(NIFTy) library 3. Learning in the Cloud to find Strong wide range of topics in artificial intelli-
Lenses). gence and the workshop fulfilled its aim.
Perhaps even further from what we usu- The available material will surely be very
ally see in astronomical conferences, useful for many years to come.

62 The Messenger 178 – Quarter 4 | 2019

Astronomical News

Demographics we had 41% students, 22% postdoctoral shop, as well as ESO catering and ESO logistics,
for ensuring the best conditions during the meeting.
researchers, and 37% tenure-track
The workshop had a very high level of or tenured faculty. The talk selection was
participation, with about 130 registered made blindly (the chair of the SOC References
participants coming from all parts of the removed names and identifying informa-
Adorf, H.-M. 1991, The Messenger, 63, 69
world and approximately two dozen tion about the authors, including their
Baron, D. 2019, arXiv: 1904.07248
unregistered participants from ESO and seniority and their affiliation), and was Griffin, R. F. 2014, The Observatory, 134, 109
neighbouring institutes, including several based solely on the merits of the abstract Stoehr, F. 2019, ASPC, 387, 523
software engineers, highlighting the and its relation to the themes of the
great interest generated by the topic. workshop. This resulted in 62% of the
talks and 50% of the posters being given
The Scientific Organising Committee by students. a
 achine learning is one of the most commonly
worked hard to ensure fair representa- used subsets of AI.
tion from the community. Among the
10 invited speakers, five were female. Acknowledgements
Three of the five sessions were also It is a great pleasure to thank the members of the
chaired by women. Among the abstracts Scientific Organising Committee (Coryn Bailer-
 orkshop website:
submitted, a quarter were by women, Jones, Henri Boffin, Massimo Brescia, Torsten­Enßlin, 2
T he Kaggle platform website:
and this was also the female/male ratio Emille Ishida, Zdenka Kuncic, Antoine Mérand, 3
Numerical Information Field Theory: http://ift.
Melissa Ness, and Felix Stoehr) for setting up an
among the contributed speakers. We amazing programme, the invited speakers for
had a very high level of participation from remarkable and clear reviews, and the organisers of
2019 AIA workshop programme: https://www.eso.
young researchers, most likely due to a the four tutorials (Patrick van der Smagt, Fabio
combination of a highly discounted Baruffa, Luigi Iapichino, Philipp Arras, Torsten Enßlin,
Philipp Frank, Sebastian Hutschenreuter, and
­registration fee for students and the fact ­Reimar Leike) for their exceptional work. We espe-
that this field is relatively young. Thus, cially thank Stella Chasiotis-Klingner for her efficient
among the registered participants, organisation of many practical aspects of the Work-

DOI: 10.18727/0722-6691/5181

Report on the IAU Conference

Astronomy Education — Bridging Research & Practice

held at the ESO Supernova Planetarium & Visitor Centre, Garching, Germany, 16–18 September 2019

Wolfgang Vieser 1 transfer from research institutes into the way to other scientific concepts, espe-
Tania Johnston 1 classroom is lacking. The goal of this cially in young people.
Saeed Salimpour 2, 3 conference was to bring together all
stakeholders — teachers, educators and Astronomy therefore plays a special role
researchers — to communicate and within public science communication.
ESO discuss their various needs in order to The literature is full of suggestions and
Deakin University, Burwood, Australia effectively bridge the gap between advice about how to best communicate
Edith Cowan University, Joondalup, astronomy education research and its astronomy to the public. Astronomy
Australia practical application. ­education or teaching astronomy is differ-
ent from communication, however.
Whereas communication and outreach
Astronomy education contributes to Astronomy is not only one of the oldest are processes aimed at generating inspi-
the spread of scientific literacy among sciences, but also a perennially fascinating ration and awareness, education aims
successive generations, helping to one to the broader public, who often ask to develop knowledge, skills and compe-
attract students into science, technology, educators questions such as, “where do tences, and core values and attitudes
engineering and mathematics (STEM) we come from?”, or “are we alone?” For through a range of pedagogies and
subjects and potentially also into astron- this reason, astronomy has always been methodologies that account for the abili-
omy research. Although the field of a relatively easy area of science to convey ties and development level of the learner.
research into astronomy education has to the public and it can serve as a gate- Astronomy education is less prominent
grown significantly, the sustainable within the scientific community than

The Messenger 178 – Quarter 4 | 2019 63

Astronomical News Vieser W. et al., IAU Conference “Astronomy Education — Bridging Research & Practice”

astronomy communication and outreach Secondary Teacher Education — span how these could be organised into coher-
even though the International Astrono­ traditional and practical research, explor- ent patterns of understanding. A new
mical Union (IAU) established Commis- ing the purely theoretical issues encoun- AER study now provides a wider and
sion 46 on “The Teaching of Astronomy” tered when attempting to embed research more coherent framework about the high
in 1964. The Commission’s designation results into practical situations, usually conceptual understanding of astrophys-
changed from 46 to C1 in 2015 but its mediated by standards, curriculum and ics that is necessary to develop research-
mandate has remained essentially the instruction. based teaching-learning sequences for
same: to further the development and high school students — something that
improvement of astronomical education This conference was organised by IAU will be developed in the near future.
at all levels throughout the world through Commission C, together with ESO, the Other contributed talks focused on how
various projects developed and main- ESO Supernova, and Leiden University. It multidimensionality in the field of astron-
tained by the Commission and by dissemi- was hosted at the ESO Supernova using omy or astronomical time- and length­
nating information concerning astronomy all its facilities, including the planetarium scales can be made understandable for
education. as the lecture theatre. The programme students. In both fields, models can help
comprised three invited talks, 44 contrib- students learn about relevant aspects,
To foster this mandate, the IAU will estab- uted talks, 10 hands-on workshops and but they need to be built by experienced
lish the Office of Astronomy for Education 50 posters. As it is an educational facility, teachers. Some contributed talks sur-
(OAE) this year; its objective will be to the ESO Supernova proved to be the veyed and analysed the production
provide structured support to for astron- ­perfect location for this conference and of AER studies in different countries like
omy education in all countries. This the participants were enthusiastic about Brazil, France, Japan and Portugal,
includes, but is not limited to, providing this inspiring environment. Details of the focusing not only on school grade levels
training and resources for encouraging programme can be found via the confer- or the type of academic research but also
the use of astronomy as a stimulus for ence webpage1. Each talk was followed on gender balance.
teaching and learning from primary to by a five-minute session dedicated to
secondary school levels. At a workshop questions and discussion that continued
between 17 and 19 December 2019 at during the breaks. Poster viewing took Astronomy education standards,
the Institut Astrophysique de Paris, the place during all coffee breaks and was ­curriculum and instruction
IAU revealed that the location of the OAE particularly encouraged during 30-minute
would be at the Haus der Astronomie in poster sessions every day. In his invited talk Robert Hollow dis-
Heidelberg, Germany. In addition, at that cussed opportunities and issues regard-
same workshop the remit of the new The IAU President Ewine van Dishoeck, ing curricula at the school level, particu-
office was presented along with its plans the IAU General Secretary Teresa Lago larly in the context of the recent IAU
regarding the goals set out in the IAU and ESO’s Director General Xavier Framework for Astronomy Literacy. The
Strategic Plan for 2020–2030. ­Barcons all acknowledged the necessity science curriculum of Australia served as
of such a conference in their welcome an example to illustrate the possibilities
The field of astronomy education has addresses. The IAU President also gave and the challenges of using astronomy in
grown significantly over the last few dec- a summary of the activities and events teaching science. A contributed talk by
ades, with an increasing number of commemorating 100 years of the IAU, Saeed S ­ alimpour gave a review of how
research articles having been published including the travelling exhibition, and often astronomy is encountered in the
by a growing number of researchers. announced the inauguration of the OAE. school curriculum of 37 countries (OECD,
Despite this, there has been no regular China and South Africa), highlighting that
international conference for astronomy 77% of all curricula in Grade 1 include
education researchers and practitioners Astronomy education research astronomy, 54% in Grades 2 & 7 and
around the world to convene and discuss 27% in Grades 1 to 12. The highest per-
their work. This conference is intended The invited talk by Janelle Bailey sum­ centage of astronomy (85%) can be
to be the first of a regular, biennial, IAU- marised the broad field of astronomy found in Grade 6. The study also revealed
Commission C1 Astronomy Education education research (AER), highlight- that one curriculum explicitly mentioned
Conference series. The aim is to increase ing upcoming projects, for example a only two women astronomers and only
the quality, quantity, community and two-volume work about astronomy three of the 37 countries explicitly men-
impact of astronomy education research ­education, and introduced modern edu- tioned indigenous astronomy.
and practice by bringing together astron- cation concepts like active learning.
omers, astronomy education researchers Future directions of AER were also dis- Several contributed talks highlighted the
and education practitioners to communi- cussed, such as the use of qualitative importance of research-based science
cate, discuss and tackle common issues. and mixed methods, robust quantitative education, in which real data are ana-
analyses and longitudinal studies. lysed with research-quality tools to inves-
The three key themes of this conference tigate questions for which the answer is
— Astronomy Education Research; Contributed talks covered more special- not known. One talk recounted how the
Astronomy Education Standards, Curric- ised topics such as students’ (mis-)con- practices employed to use archival image
ulum and Instruction; and Primary and ceptions about astronomical topics and and spectral data have evolved over time

64 The Messenger 178 – Quarter 4 | 2019

and described some of the challenges of talk about astronomy education among Figure 1. Conference participants gathered on
­balconies inside the ESO Supernova, overlooking
working with real data. The lack of user- deaf children and school-age youth in
the exhibition space called The Void.
friendly interfaces aimed at non-experts Brazil, through the Brazilian sign language
and documentation emerged as the main project Libras (Língua Brasileira de Sinais).
bottlenecks preventing the broader use Another contributed talk by Marco Brusa of time, excessively large curricula, and
of archive data. described how video games for educa- often isolation within the faculty. A prom-
tional purposes that are free of violence ising way to overcome these problems is
Another talk described an activity in and focused on STEM related science via professional development and collab-
which potential targets for the James can result in a growing interest in STEM. oration, taking advantage of the many
Webb Space Telescope are identified via European initiatives on offer. These
spectroscopic observations of stars Other internet-based education resources ­collaborations in science, technology,
taken by the Spitzer Space Telescope. were discussed like the IAU AstroEDU engineering and mathematics in general,
This activity turned out to be beneficial platform 2 for high-quality, peer-reviewed and astronomy education in particular,
not only for students, who showed an astronomy education activities, whose were intensively discussed.
increased inclination to pursue a career in Italian version was launched in 2017. The
science after this activity, but also for Open University (UK) is also accessible A contributed talk by An Steegen focused
the teachers’ levels of motivation. Another online and its ­curriculum is open to all on the level of the teacher’s awareness of
contributed talk addressed the diversity and delivered entirely by distance teach- student ideas and on the possible strate-
of curricula in a big country like Canada, ing. Its OpenSTEM Labs allow students gies they use in class related to astro-
which creates a challenge for pan-­ to perform remote experiments, including nomical concepts. Studies found that this
Canadian programmes. It was shown the use of robotic observatories. level of awareness varies considerably
that this issue can be overcome by offer- among teachers and attention should be
ing online astronomy workshops and paid to misconceptions, in both pre-­
webinars that focus on science topics Primary and secondary teacher service teacher programmes and profes-
that are common to all curricula. education sional development activities.

Internet resources like videos were shown The invited talk by Agueda Gras-­Velazquez Another contributed talk described con-
to be extremely helpful for hearing-­ focused on the struggles of teachers in tinuous professional development work-
impaired or deaf people in a contributed their daily work; challenges include a lack shops for primary and secondary school

The Messenger 178 – Quarter 4 | 2019 65

Astronomical News Vieser W. et al., IAU Conference “Astronomy Education — Bridging Research & Practice”

Australia Malaysia 38%
Austria The Netherlands 43%
Belgium Poland 57%
Brazil Portugal
Canada Romania 62%
Chile Russia 60%
China Slovakia
France Spain
Germany Sweden
Hungary Thailand
Ireland United Kingdom
Italy USA

Figure 2. Pie chart showing the distribution of Figure 3. Multi-level pie chart showing the gender
­c ountries from which the 114 participants came. ratio amongst participants (outer ring), talks (middle
ring), accepted posters (inner ring); in each of these
the lighter and darker colours represent the female
and male ratios, respectively.

teachers based around the Irish National – limitations and opportunities of plane- cated. After this meeting, the baseline
Junior Certificate theme of Earth and tariums, often seen as natural places to has now been set, and the community
Space. With these workshops, teachers run informal education activities; looks forward to marking its progress by
are kept informed of current research – how to make astronomy projects more the next conference in 2021.
and discoveries, and are provided with diverse and inclusive;
content and material to engage students – developing and testing new interdisci-
using space research. A further contrib- plinary and inclusive educational and Demographics
uted talk presented teacher trainings in outreach activities;
the use of robotic telescopes. The focus – links between astronomy and environ- The demand for this conference was
was to bring astronomy closer to teach- mental education; extremely high, but owing to the limited
ers in an enjoyable way, so that they lose – how to design inquiry-based work- seating in the planetarium, the number of
the fear of working on these topics with shops for secondary school students possible participants had to be capped
their students, and to provide them with and teacher trainings that are relevant at 114. The participants came from
the tools and knowledge so that they to curricula and cost effective. 25 countries, including 13 ESO member
can introduce them in a practical way and states, the Host Country Chile and
develop enquiry-based projects. The conference highlighted that astron- ­Strategic Partner, Australia (see Figures 1
omy education is a well-established field & 2). In total, 112 talk abstracts were sub-
with a global community. Education — mitted, 46% of which came from female
Workshops alongside research, outreach and devel- colleagues (Figure 3). The gender balance
opment — is one of the main activities among the speakers in the final pro-
The workshops covered a broad range of of the IAU. Some trends in astronomy gramme reflected the 40:60 (female:male)
activities: education recurred throughout the con- distribution of the participants, similar
– the positive and negative effects of the ference, such as multidisciplinary to that of the Scientific Organising Com-
use of technology in the classroom; approaches, the options for online collab- mittee (SOC), which had a corresponding
– the use of real astronomical data in the oration, training, and distribution. Also, ratio of 44:56 (female:male).
classroom; the societal relevance of education was
– presentation of the recently published addressed and discussed with topics
booklet Big ideas in Astronomy: A pro- like inclusion and diversity and climate Links
posed Definition of Astronomy Literacy; change. Very fruitful discussions took 1
C onference webpage:
– t wo workshops combining STEM with place during the conference and the astroedu-conference/
the arts (STEAM), one dealing with pro- majority of participants felt that they are 2
AstroEdu platform:
grammable materials that are impor- acting towards a common goal. However,
tant for future space travel, the other there is still a need to improve knowledge
with the creative use of satellite images; transfer between researchers and practi-
– an art-based approach to teaching astron­- tioners, as the wheel tends to be rein-
omy via Visual Thinking Strate­gies; vented too often, with efforts being dupli-

66 The Messenger 178 – Quarter 4 | 2019

Astronomical News DOI: 10.18727/0722-6691/5182

Fellows at ESO

Rosita Kokotanekova Rosita Kokotanekova

My path in astronomy began before I can

remember, and it has led to my becom-
ing an ESO Fellow thanks to the support
of a long list of teachers, mentors, and
friends. However, in the first place, I owe
my inspiration to be an astrophysicist
to my parents, Joanna Kokotanekova and
Dimitar Kokotanekov. They have devoted
their lives to outreach and teaching extra-
curricular astronomy classes to high-
school students in Haskovo and Dimitro-
vrad in Bulgaria.

When my brother Georgi and I were little,

our parents took us along to almost every
observation they organised: astrophoto­
graphy sessions, observations of partial After my second year at Jacobs Univer- and Akos Bogdan. This project was a
and total solar and lunar eclipses, meteor sity — in 2011 — I joined the Laboratory continuation of my work with Elke
showers, Venus and Mercury transits — of Astrophysics (LASTRO) at the École ­Roediger and would not have been pos-
you name it. Later, I participated in the Polytechnique Fédérale de Lausanne sible without her generous efforts to
Bulgarian National Astronomy Olympiad, (EPFL) in Switzerland where I worked with expand my skill set and to develop my
as well as in two International Astronomy Frédéric Courbin, Cécile Faure and resumé. My work at CfA was extremely
Olympiads in Crimea (2004) and China Georges Meylan on a six-week project to interesting and introduced me to X-ray
(2005). I also completed my first small discover strong gravitational lenses in and radio observations of galaxy cluster
research projects and had my first con- optical images from the Wide Field Cam- centres. In addition to the amazing sci-
tact with ESO, both through the Catch a era 3 (WFC3) on the Hubble Space Tele- ence environment at CfA, that summer
Star contest 1. scope (HST). I greatly enjoyed the friendly also brought me many wonderful experi-
environment at LASTRO and the Obser- ences which I shared with old and new
These experiences convinced me that vatory of Geneva, as well as living so friends in Boston.
I would like to become an astrophysicist close to the Alps, so I decided to go back
and in my bachelors degree I chose for another two-month project the year The next step of my career was deter-
to study Earth and Space sciences at after. mined by a lucky coincidence. Straight
Jacobs University Bremen, Germany. after the internship at CfA, I started look-
This programme was a great choice Straight after completing the second ing for PhD positions. While I was fasci-
because it allowed me to learn more internship, I joined the AstroMundus nated by extragalactic astronomy, and in
about geosciences and environmental ­Masters Course in Astrophysics. This particular by X-ray observations of galaxy
studies alongside astrophysics. Besides, program took me on a two-year journey clusters, I was not looking forward to
the education at Jacobs University had through four different countries, at the yet another relocation. This motivated me
a hands-on approach and prepared me University of Innsbruck, the University to keep my eyes open for other PhD
very well for a research career. of Padua, Belgrade University and opportunities that would let me stay in
­Göttingen University. After three semes- ­Göttingen or at least in Germany. Then
After only my first year at Jacobs, I con- ters of courses covering almost every suddenly, in November 2013, the press
tacted Marcus Brüggen and Elke area of astronomy, I spent the final was filled with reports about the unex-
­Roediger to ask whether I could work semester of the programme researching pected complete disintegration of comet
with them on a small research project X-ray weak quasars with Wolfram ISON. This got me very intrigued because
over the summer. During this summer ­Kollatschny at Göttingen University and up to that point I had not had any courses
project and my subsequent bachelors Luka Popović in Belgrade. This project in Solar System science and I naively
thesis research, Elke taught me a great gave me my first experience of spectros- thought that small bodies were very well
deal about galaxy clusters and hydro­ copy and taught me how to work inde- studied, and that their behaviour could be
dynamical simulations, but most impor- pendently — a skill that has come in predicted with great accuracy.
tantly introduced me to the research handy during my PhD, and especially
­process — how to start with an idea and during the ESO fellowship. That same week, a friend sent me a link
find the right collaborators, and how to to the home page of Pedro Lacerda who
complete it and produce a high-quality I had reserved the AstroMundus summer was looking for PhD students to join his
scientific publication. break in 2013 for a three-month intern- newly formed research group in Come-
ship at the Harvard Smithsonian Center tary Science at the Max Planck Institute
for Astrophysics working with Ralph Kraft for Solar System Research in Göttingen.

The Messenger 178 – Quarter 4 | 2019 67

Astronomical News Kokotanekova R., Facchini S., Hartke J., Fellows at ESO

After reading his webpage and meeting Stefano Facchini

him in person, I was captivated by
his way of thinking and his approach to
doing research. He also managed to
convince me that minor planets in the
Solar System hide many unanswered

I joined Pedro’s research group in Octo-

ber 2014 and chose to work on his large
observing programme with ESO’s New
Technology Telescope (NTT) at La Silla.
The programme was awarded 40 nights
with the ESO Faint Object Spectrograph
and Camera 2 (EFOSC2) to study the
rotational light curves and surface col-
ours of up to 60 Trans-Neptunian Objects
(TNOs). In only the third week of my
PhD, I went for my first observing run at t­hesis. At ESO, I chose to take up sup- Stefano Facchini
La Silla with my other PhD advisor Colin port astronomer duties on VLT/UT3 as
Snodgrass. This was the first time I had my functional work and now, after four Since I was a kid, I have had a passion
the chance to spend time with Colin, and shifts in Paranal, I have finally completed for science — in, I would say, two differ-
I quickly became convinced that my PhD my extensive training as a support ent flavours. First of all, I have always
was going to lead to many exciting pro- astronomer. My duties in Paranal are very been touched and fascinated by the
jects and fun trips. Soon after that run, I challenging but extremely rewarding. On beauty of nature, by the constantly vary-
enrolled as a PhD student at the Open the one hand, the trips to Chile are physi- ing shades of colour in the sea, by the
University, UK and Simon Green joined cally exhausting, but on the other hand I powerful heights of the Alps during a
the supervision team as a third advisor. have become part of the amazing Paranal hike in the summer, or by the fragility of
While most people are lucky to find community, and I have already learned field flowers in my grandparents’ farm. I
one good advisor, I was fortunate to work even more about the t­elescopes and strongly believe that my sense of awe in
with three great mentors on my PhD. instruments than I had hoped. front of the beauty and apparent order
of nature is one of the main driving forces
Like most PhDs, mine did not go as As this is the first year of my first postdoc, that led me to become a scientist. Sec-
planned. The data from the large pro- the past twelve months have been full ondly, I have always been interested in
gramme turned out to be very challeng- of many new adventures in the world of and fascinated by mathematics, show-
ing to analyse, and instead I focused research. Probably the most rewarding ing a strong propensity towards scientific
on publishing our side projects on photo- one of them was mentoring a talented topics since my first years at school.
metric observations of Jupiter-family and enthusiastic summer student —
comet (JFC) nuclei. This led to many new Abbie Donaldson — during the first ESO My passion for the night sky grew later,
ideas and accepted observing proposals Summer Research Programme (see during the first years of high school. I
on nine different telescopes. The work on page 57). For the coming year, I have an have a clear memory of one evening
that project did not always go smoothly ambitious plan, which among other being in the countryside close to Lake
either, but in the end resulted in a coher- things includes: completing a few pro- Como in Italy with a friend of mine and
ent PhD thesis, which I managed to write jects on TNOs and JFC nuclei; organising his father. His dad started pointing at
mainly during an eight-week window while the second ESO Summer Research Pro- the sky and naming the constellations
being stuck at home with a broken foot. gramme together with the other fellows; that were visible during that summer
securing more observing time for the evening. What impressed me the most is
Ever since my first trip to La Silla, I had ideas I developed over the past year; four that he had a familiarity with the beauty
been hoping to follow in Colin’s footsteps trips to Paranal; many important confer- of the sky we were looking at; he could
to become an ESO Fellow. When the time ences and meetings; more time spent recognise and name stars, whereas for
came, and I was about to look for post- working with collaborators; and last but me everything was beautiful but totally
doc positions, my advisors encouraged not least, a couple of vacations that my unknown. From that evening, I started
me to put my ideas together and design husband and I have been looking forward studying the constellations of the north-
a research programme to propose for the to for years. ern hemisphere, and I developed an
ESO fellowship application. enthusiasm for getting to know and being
able to describe the beauty of the sky.
This led to an offer from ESO and I Links
started my fellowship in November 2018, 1
 SO Catch a Star contest:
I continued to follow my passion for
two months after I defended my PhD public/outreach/eduoff/cas/ ­natural sciences, and I started attending

68 The Messenger 178 – Quarter 4 | 2019

physics courses at the University of Milan most important thing I learnt is to ask years and to even more exciting
in Italy. The choice of the subject of my myself the question that Cathie asked me discoveries!
undergraduate studies was the easiest many times: “how do you understand this
choice of my life by far! Even though equation empirically?” In other words,
I loved many topics, in particular solid how is this mathematical formula describ- Johanna Hartke
state physics and statistical mechanics, I ing a physical phenomenon in a simple
opted for a masters degree in astrophys- way? This way of looking at the mathe- It is hard to pinpoint exactly when I dis-
ics. What attracted me the most is that matical formulation of physics has covered my passion for astronomy. I grew
this subject required one to study and changed my way of doing theory forever, up in the northern German countryside,
understand many areas of physics: gen- leading me to understand a physical so even though the skies were relatively
eral relativity, classical mechanics, quan- ­process with very simple principles. dark, it was often cloudy. My parents
tum mechanics, molecular physics, etc. had a small refracting telescope which
All aspects had to be taken into account! Towards the end of my PhD, however, I stood forgotten in front of the living room
One of the topics I loved the most was felt that I was lacking something, so window, waiting for clear skies. However,
compact objects — in particular the book much so that I wondered whether to con- as a child, I was more drawn towards
Black holes, white dwarfs and neutron tinue to do research. At some point, I the piano that stood right next to it. One
stars: the physics of compact objects by understood that I was missing a closer of my first (of many) career goals was
Teukolsky and Shapiro — where the connection to observations, and I tried to to become a pianist, then followed by a
three main forces of physics interplay to find a postdoc that could allow me to desire to be a teacher, an actress,
produce beautiful objects such as neu- develop this new side of research. I was a mathematician, and eventually, a
tron stars and black holes. lucky enough that Ewine van Dishoeck physicist.
invited me to join her group at the Max
For my masters thesis, I decided to work Planck Institute in Garching, and the Following a summer school on quantum
with Giuseppe Lodato, who had recent- three years with her group have been physics for gifted high-school students
ly arrived in Milan from the UK. During my key for who I am today as a scientist. In the year before I graduated high school, I
thesis, I started working on a research particular, with her I broadened my was convinced my career lay in theo­
topic that is what I still work on seven expertise, and started working on thermo- retical physics. A year later, I enrolled to
years later: protoplanetary discs and chemical models of discs, and more study physics at Jacobs University, a
planet formation. The thesis project was directly on observations at different small, international university in Bremen. I
deeply theoretical, and we were trying wavelengths (from millimetre to ultravio- had a great experience living on campus
to answer the question: what would hap- let). Those same years, since 2015, have with students from over a hundred differ-
pen if a protoplanetary disc orbits around been transformational in my field. The ent countries, but soon realised that the-
a binary that is misaligned with respect ­tremendous capabilities of the Atacama ory was not my calling. While I enjoyed
to the disc itself? Developing semi-­ Large Millimeter/submillimeter Array experimental physics lectures, I was also
analytical models and hydrodynamical (ALMA), in terms of sensitivity and angular
simulations, we figured out that the disc resolution, together with ­high-performance Johanna Hartke
can warp, and in some extreme cases, infrared imaging instruments such as
it can break into separate annuli. At the SPHERE, completely revolutionised
time I approached this as a theoretical the field of planet formation, showing
game. How impressed I was years later images of the environments where plan-
when high-resolution images of proto­ ets form with unprecedented detail.
planetary discs started to be available, in Doing research in a field that was being
particular thanks to the VLT instruments transformed every six months by a new
NAOS-CONICA (NACO) set of observations has been among
and Spectro-Polarimetric High-contrast the most exciting experiences of my life.
Exoplanet REsearch (SPHERE), and sig-
natures of these broken discs were During the last year, I have been working
directly observed as we had predicted! at ESO as a fellow. This has allowed
me to move even more towards observa-
The masters thesis was such a great tional astronomy, getting even more
experience that I decided to keep on involved with ALMA (through my func-
doing research with a PhD. To do this, I tional work) and with other instruments
managed to go to Cambridge in the UK, on the VLT (such as SPHERE, MUSE,
to work with Cathie Clarke on a variety X-Shooter). To me ESO is the perfect
of topics, and in particular on the effects environment to do astrophysical research
that ultraviolet radiation from massive in the way I love: led by observations,
stars can have on the evolution of proto­ but with a strong theoretical background
planetary discs in young massive clus- to interpret the data and to predict what
ters. The PhD was mostly theoretical; the to expect. I look forward to the next two

The Messenger 178 – Quarter 4 | 2019 69

Astronomical News

notorious for clumsy accidents in the lab. Telescope on La Palma. During our five sis. Since the PN.S is a visitor instrument,
However, there was one topic I excelled nights at the telescope, we experienced we spent many afternoons leading up to
in and that was astrophysics. Unfortu- first-hand how it felt to be an astronomer our observations tuning the filters and
nately, the astronomy branch was closed and the patience it required in case of aligning the CCDs in the instrument arms.
in my second year of study. The subject bad weather! Yet I had found a new pas- Six months later, I got the opportunity
was not uppermost in my mind anymore, sion. It was rewarding to see our project to join my ESO Fellow mentor during his
and struggling with the prospect of grow from a little idea in our heads to duties at Paranal observatory. At last I
becoming a researcher, I seriously con- typing the coordinates of targets into the was convinced that the next step for me
sidered reverting to one of my earlier telescope, and to finally present the sci- would be an ESO Fellowship in Chile
career choices: becoming a teacher. I ence to our peers after reducing the data. to get even more exposure to the
had just made it to the state final of a One year later, I again found myself on ­forefront of astronomical research and
youth music competition in Germany and La Palma, this time observing at the instrumentation.
teaching music and physics in high ­William Herschel Telescope for my master
school seemed like the perfect combina- thesis project with Eline Tolstoy. And here I am now. I have just completed
tion of subjects for me. the first year of my fellowship and there-
It was clear that I wanted to pursue a fore the first 80 days and nights as a sup-
Everything changed, however, when I PhD in observational astronomy. In the port astronomer on Paranal. It has been
was selected for a summer internship at same year, I was accepted into the Inter- an exciting year with a steep learning
Mount Stromlo Observatory of the Aus- national Max Planck Research School curve! I am part of the Multi Unit Spectro-
tralian National University. For the first (IMPRS) on Astrophysics in Munich for a scopic Explorer (MUSE) instrument oper-
time, I got an insight into the day-to-day three-year studentship at ESO under the ations team and currently work on a
life of a researcher and could work inde- supervision of Magda Arnaboldi. For my ­project to investigate how well the adap-
pendently on a small project on stellar PhD, I investigated how the halos of tive optics improve the image quality. It
streams in the Milky Way. My supervisor early-­t ype galaxies grow through mergers is great working in an international and
Ken Freeman introduced me to the beauty and accretion. This is a challenging interdisciplinary team. I particularly enjoy
and elegance of galaxy dynamics. All endeavour, as the closest early-type the ritual of watching the sunset from
of a sudden, I could appreciate classical ­galaxies are already millions of light-years the platform before the night starts. I also
mechanics as a great tool to describe the away, but the faint halos are very recently started to experiment with astro-
motions of the stars. After the internship, extended on the sky. I therefore use a photography. I like to share the wonders
I abandoned my idea to go to the con- particular type of stars — planetary of the night sky with my friends in the
servatory and instead focused on finding ­nebulae — which are like green beacons city, where due to the bright lights, one
an opportunity to carry out my bachelor in the sky, and whose velocity can be can barely make out the Southern Cross.
thesis research project in astronomy; so measured even at a distance of hundreds When I am not observing or working
I found a placement in nearby Groningen of millions of light-years. from Vitacura, one is likely to find me
to work with Amina Helmi. rehearsing music. While living in Munich,
I enjoyed being in the middle of one of I was a soprano with the Münchner
I decided to stay at the Kapteyn Institute the astronomy hubs in Europe and got to Motetten­chor and spent a good part of
for another two years to complete my participate in many exciting seminars my leisure time in churches and concert
Master of Science, thoroughly enjoying a and conferences that were taking place halls in the region. Now in Santiago, I
curriculum centred on astronomy. Soon on campus. I travelled again to La Palma have again taken up singing, although on
an opportunity came up to enroll in a to observe the halos of giant elliptical a smaller scale. It is a relaxing balance
course on observational astronomy which ­galaxies with the custom-built Planetary to the academic world and a great way to
was to take place at the Isaac Newton Nebula Spectrograph (PN.S) for my the- practise my Spanish.

In Memoriam

ESO staff member, Cristian Herrera ator (TIO) in 2001. During his 18 years at 10 years, leading the night crew and
González, sadly passed away in August Paranal, Cristian worked on most of the was the coordinator of the Instrument
2019 and will be much missed. He joined telescopes, instruments and subsystems Operations Teams activities for the oper-
ESO and the Science Operations Depart- of the observatory. He held the role of ators during his shifts.
ment as Telescope and Instrument Oper- nighttime TIO Coordinator for more than

70 The Messenger 178 – Quarter 4 | 2019

Astronomical News

Personnel Movements

Arrivals (1 October– 31 December 2019) Departures (1 October– 31 December 2019)

Europe Europe

Andersson Lundgren, Andreas (SE) Apex Support Astronomer Gentile Bordelon, Dominic (IE) Library Technology Specialist/
Fusillo, Nicola (IT) Fellow System Administration & Classification
Girdhar, Aishwarya (IN) Student IMPRS Specialist
Heida, Marianne (NL) Fellow Harrison, Christopher (UK) Fellow
Izquierdo Cartagena, Andrés (CO) Student DFG Heijmans, Jeroen (NL) Instrumentation Engineer/Physicist
König, Pierre-Cécil (FR) Student IMPRS Hellemeier, Joschua Andrea (DE) Student
Lansbury, George (UK) Fellow Hughes, Meghan (UK) Student
Marchetti, Tommaso (IT) Fellow Iani, Edoardo (IT) Student
Oliveira Teixeira, Emanuel Pedro (PT) Accountant Kolwa Sthabile, Namakau (ZA) Student IMPRS
Paredes, Amaya (ES) Technical Writer/ Lelli, Federico (IT) Fellow
Documentation Specialist Møller, Palle (DK) User Support Astronomer
Szakacs, Roland (AT) Student IMPRS Slater, Roswitha (DE) Administrative Employee
Teuber, Karin (DE) Administrative Assistant Zanella, Anita (IT) Fellow
Trovão Ferreira, Bárbara (PT) Public Information Officer

Chile Chile

Campana, Pedro (CL) Electronics Engineer Milli, Julien (FR) Operation Staff Astronomer
Dauvin, Louise (CL) System Engineer Silva, Karleyne (BR) Operation Staff Astronomer
De Rosa, Robert (UK) Operation Staff Astronomer Vogt, Frédéric (CH) Fellow
Farias, Cecilia (CL) Telescope Instruments Operator
Hsieh, Pei-Ying (TW) Fellow
Leon, Angelica (CL) Telescope Instruments Operator
Saint-Martory, Georges (FR) ELT Deputy Site Manager
Santamaría Miranda, Alejandro (ES) Fellow
Scicluna, Peter (UK) Fellow
Slumstrup, Ditte (DK) Fellow

DOI: 10.18727/0722-6691/5183


P (selfreported | DeepThought)
DeepThought inferred knowledge


0.0047 0.22 0.78


0.1 0.35 0.55



0.52 0.37 0.11 0.15

Expert General No knowledge
G #
/ @M X lX  knowledge
- N RS G M  
@ M
+D R VD D ,NQ Referee perceived knowledge
! DS
Figure 1. Scientific seniority distribution of the DPR Figure 5. Conditional probability for the ­various
sample (blue) compared to the ESO users sample combinations of self-reported and DeepThought-­
(orange) from Patat et al. (2016). Note that the two inferred knowledge level.
central orange bars correspond to the middle
­seniority class in Patat et al. (2016).

We would like to correct and update Fig- The Messenger, 177, 3). Figure 1 is the ure 5 has been updated and corrected.
ures 1 and 5 in The Distributed Peer same as previously published and only The rest of the article, its discussion and
Review Experiment by Patat et al. (2019, has an updated caption and labels, Fig- conclusions remain unchanged.

The Messenger 178 – Quarter 4 | 2019 71