The Messenger

No. 178 – Quarter 4 | 2019 ELT M4 — The Largest Adaptive Mirror Ever Built Built Ever Mirror LargestThe Adaptive — M4 ELT GRAVITYA Celebration of Science SummerThe Research ESO Programme 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 Paranal,­ 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: [email protected] 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 Boxheimer,­ Mafalda Martins “Astronomy Education — Bridging Research & Practice” 63 Design, Production:­ Jutta ­Boxheimer Kokotanekova R., Facchini S., Hartke J. – Fellows at ESO 67 Proofreading: Peter Grimley In Memoriam Cristian Herrera González 70 ­www.eso.org/messenger/ 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/spaceengine.org

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 AdOptica/ESO 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 1 ESO opening up new discovery spaces, from Figure 1. Rendering of the M4 adaptive mirror unit exoplanets closer to their stars, to black for the ELT. 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 Zerodur©.­

Figure 3. (Right) The ­reference body in silicon-­carbide being inspected after brazing the six parts. structure being two of the most critical The back of the reference structure is 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 Boostec Mersen 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 1 Philippe Duhoux 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 1 ESO ­previous Messenger article (Kasper et al., the cold stop of VISIR to an accuracy of 2 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 3 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 4 University of Liège, Liège, Belgium tions and new technologies that were alignment of the calibration unit, which 5 Observatoire de Paris-Meudon, France implemented and how they work, con- uses an elliptical mirror with an aberration-­ 6 Subaru Telescope, Tokyo, Japan cluding with the execution of the NEAR free field of view of around 0.1 mm in 7 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 8 Caltech, Pasadena, USA vation of the a Centauri system, which mounts on-site. 9 Uppsala University, Sweden took place in early June 2019. 10 Laboratoire d’Astrophysique Marseille, Following the completion of the assembly France Three years were needed to develop the integration and verification (AIV) activities, 11 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- 12 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- 13 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 Planets­ in a Cen

the instrument, and some fine adjustment Figure 1. (Left) VISIR of the wavefront sensor arm. On-sky mounted on UT4 and 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- tested, and operational procedures were Collaboration ESO/NEAR tronics racks and helium compressors on the tuned. ­azimuth platform can be seen on the left hanging down from the mirror Technical innovations, observing modes cell. 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 Chopping, internal and external acquisition of a Centauri with chopping. Among the new technologies is an internal chopping device, the so-called Dicke Switch, which is described in more detail a Cen B a Cen B in Kasper et al. (2017). We tested the Dicke Switch at chopping frequencies up to 10 Hz during commissioning, and a Cen A – a Cen B 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

D x λ /

( 8 y 100 or

0.05 7 ) err (h

) 6 80 (h

ing me 0.00 5 ight

60 ve ti point /n 4

ed –0.05

Time 3 40

2 Cumulati

Estimat –0.10 20 1 0 0 ) 0.25 12345678910 11 12 13 14 15 16 17 18 19 20 21 Nights of 23 May–11 June, 26 June Time/night (h) Cumulative time (h) seconds 0.20 rc (a Figure 3. (Left) Top: The shifting move- Figure 4. (Above) Hours of open shutter time over

ion 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 posit

a closed loop during a NEAR observ- 28 May and between 5 and 11 June, owing to medio-

or 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, select centered on meridian passage, the rms estimation is 0.015 λ/D for both Figure 4 shows the campaign progress in field the x and y directions. Bottom: The data hours collected per night. The maxi- ve 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

Relati 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, ­QACITS 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 Planets­ in a Cen

Separation (λ/D) Figure 5. Left: The deepest ever 0510 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. )

σ Right: Sensitivity and contrast 5 ­estimated from the deep image as y, –5 (J –3 10 a function of radial distance to the 10 rast

ty centre.

Affected strongly Cont by ADI residuals Sensitivi

10 –4 10 –6

0246810 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 shows some straight lines connecting the the achieved signal-to-noise ratio with the for access to the Infrared Lab at ESO, Paranal’s 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- sistence in the detector, i.e., the pixels Centauri campaign are publicly available, ful discussions and their interest in the data analysis. α 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 may hide another 5–10% of the possible 3 λ/D and the N-band sensitivity are during the rapid development of the experiment. planet orbits. unprecedented in ground-based astron- omy by a large margin — more than one References order of magnitude. The sensitivity limits Beyond NEAR are well understood and could be Anglada-Escudé, G. et al. 2016, Nature, 536, 437 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 substantial interest within the community cold pupil stop in front of the AGPM. But Lagage, P. O. et al. 2004, The Messenger, 117, 12 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 Links tion and resulted in 26 proposals being sensitivity once more. These next-gener- 1 Breakthrough Initiatives webpage: http://break- submitted for NEAR observing time. Two ation detectors would allow the VLT to throughinitiatives.org periods of Science Demonstration were probe the rocky planet regime in the hab- 2 SPARTA: https://www.eso.org/sci/facilities/ allocated in September and December itable zone around Centauri. When develop/ao/tecno/sparta.html a 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 ([email protected]) or Markus Kasper posed programmes. other southern hemisphere 8-metre-class ([email protected] ESO/NEAR Collaboration ESO/NEAR

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 1 ESO and Infrared Survey Telescope for Astron- under the leadership of the PI. The agree- 2 Terma GmbH, Darmstadt, Germany omy (VISTA) required them to span a time ment states that a final release which 3 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. s) 100% ESO Public Survey status 80% A total of twenty Public Surveys1 have etion (OB hour been carried out by consortia in the com- 60% munity and are actively supported by ESO. The majority have completed data compl acquisition using ESO facilities and are of 40%

in the process of publishing science data tage products via the ESO archive. en 20% rc ATLAS

Pe KIDS ESO Public Surveys were launched in 0% VPHAS+ 2005 with an initial call for the optical 8 2 4 8 2 4 8 2 4 8 2 4 8 2 4 8 2 4 8 2 4 8 2 4 imaging surveys at the VLT Survey Tele- -0 -1 -0 -0 -1 -0 -0 -1 -0 -0 -1 -0 -0 -1 -0 -0 -1 -0 -0 -0 -0 -1 -0 11 11 12 12 12 13 13 13 14 14 14 15 15 15 16 16 16 17 17 17-118 18 18 19 scope (VST; Capaccioli & Schipani, 2011), 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 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% s) 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 etion (OB hour VMC VIKING for Cycle 2 VISTA imaging Public Surveys 60% was opened in 2015 and the selected VVVx compl surveys began in April 2017 (Arnaboldi et G-CAV of al., 2017). Four of the seven Cycle 2 VISTA 40% VEILS surveys exploit the time domain: for tage SHARKS example, following up exotic transients en 20% UltraVISTA-New rc VISIONS

like the optical-near-infrared echo of grav- Pe itational wave (GW) events (VinRouge); 0% VinRouge 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 0 2 6 studying the 3D shape of the Milky Way -1 -1 -0 -1 -0 -1 -0 -1 -0 -1 -0 -1 -0 -1 -0 -0 -1 -0 -0 0-0 bulge (VVVX) via astrometry — to test 09 1 10 11 11 12 12 13 13 14 14 15 15 16 16 17 17-118 18 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 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 supernovae (SN) in cosmological deep tion of the VST surveys. The VST ATLAS survey was for Cycle 1 and 2 VISTA Surveys. The cumulative 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- surveys, very much in the spirit of other it be terminated early. sation via reports submitted to the OPC. 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 1350b u’ g’ r’ i’ 24.1 24.6 24.4 3421 http://kids.strw.leidenuniv.nl/ 23.4 (de Jong et al., 2013) ATLAS Wide area/baryon 4700c u’ g’ r’ i’ z 22.0 22.2 22.2 1585 http://astro.dur.ac.uk/Cosmology/vstatlas/ acoustic oscillations 21.3 20.5 (Shanks et al., 2013) VPHAS+ — VST Photometric Hα Survey of the Southern Stellar astrophysics 1800d u’ g’ Hα r’ i’ 21.8 22.5 21.6 1200 Galactic Plane 22.5 21.8 http://www.vphas.eu (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 http://home.strw.leidenuniv.nl/~ultravista/ 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 http://www.ast.cam.ac.uk/~rgm/vhs/ 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 https://www.eso.org/sci/observing/PublicSurveys/ 24.0 23.5 sciencePublicSurveys.html (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 http://vvvsurvey.org/ 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 http://www.astro-wise.org/ 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 http://star.herts.ac.uk/~mcioni/vmc/ 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 http://www.star.le.ac.uk/nrt3/VINROUGE/ 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 http://home.strw.leidenuniv.nl/~ultravista/ VVVX* — Extending VVV to higher Galactic latitudes Milky Way 1700 J H Ks Ks = 17.5 1631 http://vvvsurvey.org/ VEILS* — VISTA Extragalactic Infrared Legacy Survey Galaxy evolution, 9 J Ks J < 23.5 847 http://www.ast.cam.ac.uk/~mbanerji/VEILS/veils_index.html AGN, SN Ks < 22.5 G-CAV — Galaxy Clusters At VIRCAM Galaxy clusters 30 Y J Ks 24.5 24 23 440 http://www.oats.inaf.it/index.php/en/2014-09-12-12-59-22/ tematiche-di-ricerca/macroarea-1-en/670-galaxy_cluster.html VISIONS* — VISTA star formation atlas Star formation 550 J H Ks 21.5 20.5 19.5 449 https://visions.univie.ac.at atlas SHARKS — Southern Herschel-Atlas Regions Near-infrared 300 Ks 22.7 929 Ks-band survey counterparts for https://www.iac.es/sharks/ 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 http://www.gaia-eso.eu/ populations (Randich et al., 2013) PESSTO — Public ESO Spectroscopic Survey of Transient Objects Transient, 150 ~ 2500 384.0 http://www.pessto.org/ SN progenitors (Smartt et al., 2013) VANDELS Physics of galaxies in the 2700 ~ 1500 142.7 http://vandels.inaf.it early universe CANDELS, (McLure et al., 2017) UDS & CDFS fields LEGA-C — Large Early Galaxy Astrophysics Census Dynamics of galaxies 3100 ~ 1500 99.8 http://www.mpia.de/home/legac/index.html 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 s completed in 2017. The surveys using the VIsible 600 Refereed publications Multi Object Spectrograph (VIMOS), called

at ion Citations (since 2010)

­VANDELS and LEGA-C, began operations in 2015 it Total ref. publications: 848 and were completed in March 2018, before the /c 500 172 (20.3%) from archive decommissioning of the VIMOS spectrograph. 400 84 (10%) are archive + PI independent archives (for example, the

ublications 300 VISTA science archive, Vizier) or the

­Public Survey webpages have also been of p 200 made available to those in the community interested in accessing data products for 100 their independent scientific explorations. Number 0 S O C X S S C TA V/ SO TO IS VH LA GA The ESO library routinely monitors VM HAS+ KIDS –E SS aV VIDE VIKING VV AT NDEL LE VP PE ­refereed publications, based on data Gaia VA VINRouge Ultr VST 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- as well as archive papers. Archive papers ereed publications and citations (divided by 10) for 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 News­letter9 bers of queries carried out using the ESO of their science data through the Phase 3 and the ESO archive community forum10. catalogue query interface12 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 Survey science products published via releases from the ESO Public Surveys published by 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

VPHAS+ DR4 VEILS DR1 V-ATLAS DR4

PSF Phot. MW DR1 VMC DR5 KIDS DR4

VINROUGE DR1 VISIONS DR1 G-CAV DR1 UltraVISTA DR4

14 The Messenger 178 – Quarter 4 | 2019 18 000 250 000

16 000

14 000 200 000

b) 12 000

files 150 000 e (G 10 000 of lum 8000 vo 100 000

6000 Number Data

4000 50 000 2000

0 0 2 2 3 1 2 2 3 3 3 7 7 4 6 7 7 8 4 4 4 6 8 8 8 6 6 9 9 9 8 9 8 7 7 6 6 3 4 4 1 3 2 2 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –15 –1 –15 –15 –15 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 –1 5 –15 –1 –1 –1 –1 –1 –1 12 12 12 12 12 12 12 12 12 12 12 12 12 12 03 12 12 03 03 09 06 03 03 03 09 03 03 06 09 06 09 06 09 09 06 06 09 09 06 06 03 06 06 09 06 06 06 06 VHS VVV/3 UltraVISTA VMC VIDEO VHS VVV/3 UltraVISTA VMC VIDEO VIKING VPHAS+ KiDS V-ATLAS Gaia-ESO VIKING VPHAS+ KiDS V-ATLAS Gaia-ESO PESSTO Lega-C Vandels G-CAV VISIONS PESSTO Lega-C Vandels G-CAV VISIONS data products from ESO Public Surveys Figure 5. Charts show- 2500 is 46% of the total number of users ing the cumulative data volume (upper left) and (3457) who have downloaded science numbers of files down- 2000

data products from the ESO Archive. loaded (upper right) ueries

for each ESO Public /q Survey. The lower chart sers

shows the cumulative /u 1500 How ESO promotes Public Survey curves for the number of science catalogues (× 10), the number of distinct talogues 1000 ESO promotes science results from users, and number of ca queries (× 0.1) related to of ­Public Surveys via science/photo press the ESO Public Surveys releases and community workshops. downloaded using 500 umber

The most recent photo release of the new the ESO catalogue N image of the Large Magellanic Cloud14 query interface. celebrated the 500th anniversary of its 0 12 13 13 14 14 14 15 15 16 16 17 17 18 18 19 first being seen by Europeans (during p– r– t– r– t– c– n– c– n– c– n– c– n– c– p– the voyage of the explorer Ferdinand Se Ma Oc Ma Oc De Ju De Ju De Ju De Ju De Se 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 support the scientific use of the available for 4MOST Community Surveys on Edge, A. et al. 2013, The Messenger, 154, 32 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 The Public Survey Panel review on VISTA over a period of five years. McLure, R. et al. 2017, The Messenger, 167, 31 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 are described in the special issue of the Sutherland, W. et al. 2015, A&A, 575, 27 7 Tanvir, N. et al. 2017, ApJ, 848, 27 In 2005, the Public Survey Panel (PSP) Messenger in March 2019 . 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 Links 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 http://www.eso.org/sci/observing/PublicSurveys/ of the panel: Duccio Macchetto, Simon characterised by large investments that sciencePublicSurveys.html 2 White, Danny Lennon, and Miguel Mas comprise dedicated telescopes and ESO Telescope Bibliography telbib: http://telbib.eso.org 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 http://casu.ast.cam.ac.uk/ 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 http://www.roe.ac.uk/ifa/wfau/ 5 TERAPIX: http://terapix.iap.fr/ and VST Panels were later merged. Since of such survey projects includes the leg- 6 CALE T: https://calet.org/ 2018 there has been only one panel for acy value of science products that are 7 Astro-WISE: http://www.astro-wise.org/ all ESO Public Surveys, whether imaging made available through the archives for 8 ESO Phase 3 Data Releases: http://eso.org/rm/ or spectroscopy. In the near future, the further scientific exploration by the com- publicAccess#/dataReleases 9 ESO Science Newsletter: http://www.eso.org/sci/ 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 https://esocommunity.userecho.com/ 11 scientific advancement of its community. ESO Archive Science Portal: https://archive.eso.org/scienceportal/ The PSP review in May 2019 12 ESO catalogue query interface: https://www.eso.org/qi/ As most of the Public Surveys completed Acknowledgements 13 ESO Archive Programmatic Access webpage: their data acquisition in 2019, the goal http://archive.eso.org/programmatic/ 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- https://www.eso.org/public/news/eso1914/ 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 https://www.eso.org/public/news/eso1339/ 16 Surveys using criteria that included pub- from the ESO Directorate of Engineering who sup- VISTA image from KIDS survey: ported the development of the tools required for car- https://www.eso.org/public/news/eso1642/ 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 https://www.eso.org/public/news/eso1733/ tific extensions, complementarity with interfaces. The authors thank the ESO library team 18 Messenger issue dedicated to 4MOST GTO other telescopes, activities to promote for their careful monitoring of the refereed publica- ­surveys: https://www.eso.org/sci/publications/ 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 5 Public Surveys were invited to the review TERAPIX , for their hard work and support of the Notes 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 Astrofit2 Marie Skłodowska-Curie Fellow during their presentations. All PIs but one served as members in the different PSPs for their b The 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 a lively and constructive discussion. The to execute the survey programmes. footprint. c The 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 The 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 While work on the scientific analysis con- Cioni, M.-R. et al. 2013, The Messenger, 154, 23 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 2, 3 Elena Dalla Bontà include the Elodie library (Soubiran et 2 2, 3

Enrico M. Corsini al., 1998; Prugniel & Soubiran, 2001; g 4

Lorenzo Morelli Le Borgne et al., 2004) and the X-shooter og Spectral Library (XSL; Chen et al., 2014). The latter showcases the problems that 1 ESO increasing resolution and multi-order 4 2 Dipartimento di Fisica e Astronomia cross-dispersed spectrographs bring; “G. Galilei”, Università di Padova, Italy synthetic broadband optical (UBV) colours 3 INAF–Osservatorio Astronomico di show poor agreement with observed Padova, Italy ­colours from the Bright Star Catalogue 4 4.5 4 3.5 Instituto de Astronomía y Ciencias (on average at ~ 7%; see Table 5 and Fig- log T ­Planetarias, Universidad de Atacama, ure 26 in Chen et al., 2014). The differ- eff 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 Nl 2 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 AFG KMC/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 T for stars with of our spectra with synthetic broadband troscopic Explorer; Bacon et al., 2010) eff [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

12

10 HD 057060 07e

10 t) t) HD 096446 B2III 8 HD 174966 A3

HD 167278 F2 s + cons s + cons unit

unit 6

HD 163810 G3V

ry ry ra

ra HD 167278 K3V rbit rbit 5 (a (a

4 λ λ F F (B86) 133 M4

2

0 HD 067507 CNv 0 6000 8000 104 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 ratios and extrapolated them over the full DOR1885254/18, DOR1935272/19, and 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 noise ratios S/N > 70–200 and are availa- Freudling, W. et al. 2013, A&A, 559, A96 2 Ivanov, V. D. et al. 2019, A&A, 629, 100 ble via the ESO MUSE webpage or via Finally, we calculated synthetic Sloan Le Borgne, J.-F. et al. 2003, A&A, 402, 433 3 CDS/VizieR . 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 range covered by MUSE were measured Ivanov et al., 2019). The MUSE sequences Prugniel, P. & Soubiran, C. 2001, A&A, 369, 1048 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 The XSL library: http://xsl.u-strasbg.fr/ 2 between the 6 (or 12 in the case of some of the colours can mix fluxes from The MUSE spectral library at the ESO MUSE web- page: https://www.eso.org/sci/facilities/paranal/ 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 The MUSE spectral library at VizieR/CDS: http:// direct comparison of the MSL and XSL cdsarc.u-strasbg.fr/viz-bin/cat/J/A+A/629/A100 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 spectral order — is excellent. However, astronomiques Strasbourg (CDS) and of the VizieR 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 Enrico GRAVITY Science

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 Gillessen1 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 Anugu7,29,30 Rebekka Grellmann 4 André Müller 3 Gerardo Ávila 8 Ulrich Grözinger 3 Eric Müller 8, 3 Michi Bauböck1 Patricia Guajardo 8 Friedrich Müller 3 Myriam Benisty 5,12 Sylvain Guieu 5 Hagai Netzer 32 Jean-Philippe Berger 5 Maryam Habibi1 Udo Neumann 3 Joachim M. Bestenlehner 22, 3 Pierre Haguenauer 8 Mathias Nowak 2 Hervé Beust 5 Oliver Hans1 Sylvain Oberti 8 Nicolas Blind9 Xavier Haubois 8 Thomas Ott 1 Mickaël Bonnefoy 5 Marcus Haug 8 Laurent Pallanca 8 Henri Bonnet 8 Frank Haußmann1 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 Buron1 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 Rosales1 Jaime E. Pineda1 Benjamin Charnay 2 Lieselotte Jochum 8 Philipp M. Plewa1 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 Davies1 Stefan Kellner1 Sebastian Rabien1 Casey Deen1 Sarah Kendrew 11, 3 Andrés Ramírez 8 Françoise Delplancke-Ströbele 8 Lothar Kern8 José Ricardo Ramos 3 Roderick Dembet 8 Pierre Kervella 2 Christian Rau1 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 Kok1 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-Cox1 Pierre Fédou 2 Vincent Lapeyrère 2 Jinyi Shangguan1 Miguel C. Ferreira 6 Werner Laun 3 Thomas T. Shimizu1 Gert Finger 8 Bernard Lazareff 5 Jason Spyromilio 8 Natascha M. Förster Schreiber 1 Jean-Baptiste Le Bouquin 5 Amiel Sternberg1, 32 Feng Gao1 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 Sturm1 Éric Gendron 2 Magdalena Lippa1 Marcos Suárez Valles 8 Reinhard Genzel 1,15 Dieter Lutz1 Linda J. Tacconi1 Ortwin Gerhard1 Yves Magnard 5 Wing-Fai Thi1

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 Dishoeck14 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 Fellenberg1 de México, Mexico Idel Waisberg1 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 Widmann1 Munich, Germany ity atomic gas near an accreting super- Ekkehard Wieprecht1 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 Wiezorrek1 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. 26 Space Telescope Science Institute, We will measure a new radius-luminosity 1 Max Planck Institute for Extraterrestrial Baltimore, USA relation from spatially resolved data and Physics, Garching, Germany 27 School of Physics & Astronomy, test the current methods used to meas- 2 LESIA, Observatoire de Paris, Université ­University of Southampton, UK ure black hole mass in large surveys. PSL, CNRS, Sorbonne Université, 28 Department of Astronomy, California ­Université de Paris, Meudon, France Institute of Technology, Pasadena, 3 Max-Planck-Institut für Astronomie, USA Introduction Heidelberg, Germany 29 Steward Observatory, Department 4 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 Univ. Grenoble Alpes, CNRS, IPAG, 30 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 CENTRA and Universidade de Lisboa 31 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 CENTRA and Universidade do Porto – 32 School of Physics and Astronomy, observed properties can be explained Faculdade de Engenharia, Porto, Tel Aviv University, Israel by emission from discrete, collapsed Portugal 33 Excellence Cluster Origins, Ludwig- clouds or high-density regions of a con- 8 ESO Maximilians-Universität München, tinuous medium. The gas may be part 9 Observatoire de Genève, Université de Garching, Germany of the inflow feeding the black hole or a Genève, Versoix, Switzerland 34 Universitäts-Sternwarte München, continuous equatorial outflow. Assuming 10 DOTA, ONERA, Université Paris- Munich, Germany a gravitational origin, line widths com- Saclay, Châtillon, France 35 Shanghai Institute of Optics and Fine bined with a measurement of the emis- 11 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 36 NOVA Optical Infrared Instrumentation 12 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, 37 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 38 Observatoire de la Côte d’Azur the sample size from tens (Kaspi et al., 13 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 Sterrewacht Leiden, Leiden University, 39 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 flux α m) 0.6 2.176 (µ 1.4 ) )

10 th (µas x

e (° 0.4 1.2 2.174 leng on flu ti 0 phas 1.0 2.172 wave

al 0.2 ti ed

en 0.8 –10 rv Δ Declina er 2.170

0.0 Normalised ff

Di 0.6 Obse Phase –20 –0.2 2.168 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° Y 100 2 )

/ –1 s

) 50 1 km

ev e¦ 3 (µas

H 0 (1

 ion "KNTC 0 0 loci ty e¦

v ve , e %+ –50 –1 MFTK@Q Δ Declin at BKNTC

CDMRHSX Radial –100 –2

1@CH@K ` BKNTC CDMRHSX –150 –3 150 100 50 0 –50 –100 –150

1LHM 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 –1 another detection and upper limits in deep integra- c) 10 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 Radius (p ­(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 10 43 10 44 10 45 10 46 10 47 velocity structure well ordered or randomised? 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 + 9 (Rakshit et al., 2015) whose detection angle of 45–6 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 8 tions. This is now possible with GRAVITY. 1.5–4.1 × 10 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 at wavelength channels where the line we are carrying out observations of CEA and CNES, by the Programme National GRAM 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 entation perpendicular to the large-scale information on the dominant kinematics de Physique Stellaire (PNPS) of CNRS/INSU, co- 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 ical precision of 5 µas per channel. ing questions: are the line widths pri­ the European Research Council (ERC) under the

The Messenger 178 – Quarter 4 | 2019 23 GRAVITY Science

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 15 Maser disc −300, 0, 300 km s–1 Po 10 larise d ou tfl ow

) cone

as 5 (m et fs

of 0 −1

(X-ray) NASA/CXC/MIT/C. Canizares, Evans D. et al., (optical) NASA/STScI, (radio) NSF/NRAO/VLA (mJy beam )

at ion 40. −5 Declin 30.

20. −10 10. 1 pc −15 0. 1600 pc 15 10 50−5 −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 angle of 50 degrees west of north, an an additional test, we have compared the optical emission is shown in green, the X-ray in red 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, radius for large graphite grains in the ranges, the models tend to over-predict fitted to the brightest knots, traces a ring that 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 45 bolometric luminosity of ~ 4 × 10 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 aligns the central hole in the near-infrared As an alternative, we considered whether derived from the maser kinematics, and hence fixes 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 then the positions of the lower-velocity vides an unexpectedly good fit to the polarisation data. The grey dashed ellipse indicates 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 GRAVITY Science

expected for the dust sublimation region, and the apparent orientation is similar ° to that of the maser disc, arguing for 70 a common origin. The structure and pho- i = tometry are consistent with dust at ight ~ 1500 K behind AK ~ 5.5 magnitudes of s foreground extinction. This matches what of is expected from the upper limit to the Line 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- tinuum would originate in a separate r clouds structure, likely associated with the AGN- Mase Thick molecula r gas disc driven outflow.

Acknowledgements c 5 dis in t th See page 23. Ho 1 dust c) MID-IR (p References 25 R 0. 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 Y. 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 Chile to form the VLT Interferometer With GRAVITY and SINFONI we were laser of Unit Telescope 4 (Yepun) points at the 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 diameter (GRAVITY Collaboration, 2017). absorption lines in the light of S2 were the centre, which harbours the (invisible) 4 million 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

4000 0.175 3000 )

–1 2000 0.15 s m (k

1000 R LS v 0.125 0 )

ೀ –1000 ( 0.1 –2000 at ion 2000 2005 2010 2015 2020 0.075 Time (year)

Δ Declin 250 0.05

) 200

0.025 –1 s m

(k 150

0 R LS v

100

0.05 0.025 0 –0.025 –0.05–0.075 50 Residuals 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. 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 ) emission becomes a bright flare, and at times it contains so much energy that (µas it even emits X-rays. The true nature of et 0 these flares seen at infrared and X-ray fs of

7 Rg Calçada Consortium/L. ESO/GRAVITY image: Background wavelengths is not yet known and may y- be explained as a hot spot in the gas or an ejected blob of gas (as in a jet).

Observations during the summer of 2018 –100 Figure 3. Projected orbit 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 centre (GRAVITY Collaboration, 2018b). to dark blue indicates 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 GRAVITY Science

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 101102 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

2.50 a) Brδjet1 Brγ Brγjet1 Brγjet2 JeJ 2.25 t in He I He I He I te jet1 jet2 N nsitns 2.00 it y pr co nt 1.75 ofiofile F

)/ le λ (

F 1.50 E - - - - 1 1.25

1.00 1.7 m 7 ma 1.2 b) ass

y UT1–UT3 - - - 1/e isibilit

d v 1.0 Normalise 0.8

) c) ees 10 0.26 c

egr UT1–UT3 s (d

hase 0 1mas d p

–10 > 15 mas Normalise

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- –5 –1 < 0.5 astronomical units [au]), and (ii) simple geometric models. ful winds (~ 10 M⊙ yr ) 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 GRAVITY Science

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

0.4 0.4 Brγ stationary ) ) –1000 0 1000 as as Brγ stationary 0.2 v (km s–1) 0.2 (m –1000 0 1000 –1 ion v (km s ) Approaching jet t at io n (m 0.0 0.0 Declina Declin Approaching jet –0.2 –0.2 Receding jet

–0.4 0.2 mas ≈1au –0.4 0.2 mas ≈1au 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. ­emission 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 with a size several times the stellar radius, orbital phases could take advantage of GRAVITY Collaboration et al. 2017b, A&A, 602, L11 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 could be caused by the X-ray ionisation accretion. Waisberg, I. et al. 2019a, A&A, 623, A47 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 –1 –1 –1 –1 of the “Homunculus –693.0 km s –657.0 km s –620.0 km s –584.0 km s 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 ) –1 –1

as 3

514.0 km s 551.0 km s x 10 flu (m 2 m. ion 0 at

Nor Wavelength (m) –10 1 2.162 2.164 2.166 2.168 2.170 Declin 1e–6 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,

as 10 2018) at a Doppler (blue-shifted) velocity of –280 km s–1. The projected trajectory of the second- (m 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. Milliarcseconds –15 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 decisive­ 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 ­Multi-BEam combineR (AMBER) obser­ spectral channels, with a resolution as Clementel, N. et al. 2015a, MNRAS, 450, 1388 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 wind-wind collision cavity across Br tural changes associated with the orbital Mehner, A. et al. 2012, ApJ, 751, 73 g 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 Observing Car with GRAVITY identified as part of shocked wind flowing Thiébaut, E. 2008, Proc. SPIE, 7013, 701311 h 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 Wittkowski1 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 Chiavassa3 medium via stellar winds. ­Yasuda & Kozasa, 2019; Rau et al., 2019). Willem-Jan de Wit1 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 Kravchenko1 shown that pulsation and convection lead Steiner et al., 2014), but an application Claudia Paladini1 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 1 ESO (for example, Höfner & Olofsson, 2018). Time-series of interferometric observa- 2 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 3 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- 4 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 5 NASA, Goddard Space Flight Center, pulsation and convection alone, cannot time-series are still very rare. Greenbelt, USA explain the observed extensions of RSG 6 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- 7 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 0.0 0.2 0.4 0.6 0.8 crosses) and of the uniform disc angular diameter in a near-continuum band (blue × symbols), and in bands dominated by H O and CO (light blue and 4 R Peg 2.25 µm (Cont.) 2 14 pink × symbols, respectively). Also shown are sinu-

VLTI-GRAVITY 2.05 µm (H2O) soidal fits in the corresponding colours. The mini- 2.29 µm (CO 2-0) ) mum continuum size tracks the maximum light, 6 as which can be understood by the increase in effective

(m temperature while the star gets smaller in radius. 12 The minimum contribution of H O also tracks the

er 2 e 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 sc diamet magnitud bly more stable, behaviour compared to H2O. From 10

V Wittkowski et al. (2018). m di or

12 8 Unif

14 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 observed amplitude of the continuum Combined with spectro-interferometric Kravchenko, K. et al. 2018, A&A, 610, A29 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 dicted by 3D RHD simulations (Freytag acterisation of AGB and RSG star atmos- Wittkowski, M. et al. 2018, A&A, 613, L7 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 GRAVITY 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: http://www.mpe.mpg.de/ir/ gravity Outlook Acknowledgements 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 consortium2, the GRAVITY Collaboration (see which current models successfully pre- page 20), and the ESO science operation team for 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 References 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 sons to the latest dynamic models. Bladh, S. et al. 2015, A&A, 575, A105 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 our pilot study by applying a tomographic GRAVITY Collaboration 2017, A&A, 602, A94 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 1 mode of massive star formation. One of for q 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 forn 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 1 sive object. q 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 total of 22 companions. The scale of B5 B1 B 4 the separation of the companion is indicated in the figure. The coloured B images of 1 Ori B are from observa- 0.1 au 1, 5 q tional data, except the greyscale B1, B 1, 5, 6 B5 system, which is only representa- NU Ori tive. The image of B1, 5 and B6 is a 5 au B6 reconstructed image of GRAVITY B2 0.3 au observations. The orbital positions, B 3 which are indicated for q1 Ori C and q1 Ori D, are the positions given in GRAVITY Collaboration (2018) and 2 0.1 au 157 au θ Ori A ­previous 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 B E 100 au θ2 Ori C D A 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)

observed 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 ) ๬

) Nu Ori (16.01 ± 3 M ) ๬ ๬ stars M (

1 of θ Ori D (16.0 ± 1.0 M ) ๬

2 mass θ Ori B (14.8 ± 3.4 M ) ๬ y Number θ1 Ori A (14.0 ± 5.0 M ) ๬ 10 –2 θ1 Ori B (7.2 ± 0.2 M ) ๬ & primar

em HD 37115 (5.4 ± 0.4 M ) ๬

Syst 2 1 θ Ori C (4 ± 1.0 M ) 10 ๬

Mass (M ) 1 ๬ 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 0 101 102 Companion separation (au)

The Messenger 178 – Quarter 4 | 2019 37 GRAVITY Science

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, mergers as the dominant formation main mechanism for the formation of high IAU Symposium, 200, 346 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 Summary & conclusions crucial role of interferometry in filling the Grellmann, R. et al. 2013, A&A, 550, 531 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 between 1 and 100 au. Almost all massive Tan, J. C. et al. 2014, Protostars and Planets VI, ed. 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 GRAVITY1 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) Closest Evolution of the inner structure

o HD 97048 (g) HD 97048 (g) ati

HD 38120 HD 38120 to th An underlying question related to disc ze r si classification is whether the sources K

- HD 144432 (g) 10.0 HD 259431 e with flared/gapped and flat/continuous -to HD 150193 HD 150193 ZA N HD 158643 discs form an evolutionary sequence.

HD 142527 (g) MS HD 179218 (g) ­Dullemond and Dominik (2004) proposed HD 139614 (g) HD 95881 HD 163296 HD 163296 that discs might start out with a flared HD 98922 HD 45677 HD 144668 HD 144668 shape, then become flat when going HD 135344 (g) 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 1.02.0 5.010.0 primordial flared discs. More recently, Star mass (M ) ๬ 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 posed that they could indicate a gapped by ESA Gaia observations (Vioque et al., the 2 M⊙ objects and the arrow indicates the position 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, theK -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-­ luminosity­ relationship. For the luminous 3 FUV/EUEUVV//XX--rraay stars (around 10 L⊙), a large scatter 10.0 radius around the mean relation is observed, Figure 2. K-band emis- HD 45456677 Critical pointing towards a range of compositions sion location as meas- ured by GRAVITY as a of the inner dusty discs. u)

(a 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. -band emission location 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 mid-Infrared SpectroScopic Experiment ets and EUV/X-ray (MATISSE 3) can now investigate disc ­photoevaporation will open gaps and not ­flaring at tens of au and can help to ques- 1.0 2.05.0 1100.0.0 20 deplete the inner disc tion disc classifications based on Spec- Star mass (M ) quickly (bottom inset). ๬

The Messenger 178 – Quarter 4 | 2019 39 GRAVITY Science

Ae/Be mass range and need additional emission we measured with GRAVITY is GRAVITY Collaboration et al. 2019, A&A, 632, A53 observations. located at positions smaller than the Le Bouquin, J.-B. et al. 2011, A&A, 935, A67 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 sample might be shaped by forming Vioque, M. et al. 2018, A&A, 620, A128 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 GRAVITY 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 and far-ultraviolet (EUV/FUV) and X-ray and spectral resolution. 1.8 µm. 3 MATISSE 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 6 takes a few 10 years and inner disc Beuzit, J.-L. et al. 2019, A&A, 631, A155 depletion takes about 105 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 1.25 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. uum. There is a strong differential phase 2–03–1 4–25–3 1.20 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

x Berthoud et al., 2007). Near-infrared indication of an asymmetric circumstellar

Flu 1.10 interferometric observations using the environment. Astronomical Multi-BEam combineR (AMBER) — one of the first-generation By fitting our interferometric observations 1.05 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° –1 10 (v sin i ~ 147 km s ) and located at roughly 0.1 au from the star (more details The circumstellar environment of in Koutoulaki et al., in preparation; see phase 5 51 Oph also the spatial scales probed in Fig-

ntial ure 3). From the model, an intensity map re 51 Oph was observed with GRAVITY could be created since we know the fe if 0 ­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 eD The differential phase is the photocentre observations. By varying the inclination

phas 5 shift of the line with respect to the con­ and position angle the software creates ial t tinuum and the closure phase is an indi- synthetic observations that can be com- en

er 0 cator of asymmetries in the circumstellar pared with the real ones. ff

Di environment. –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 –10 The spectrum shows four bright band heads (u = 2–0, 3–1, 4–2 and 5–3). The 2.30 2.35 2.40 Figure 2. Intensity maps of the blueshifted (left), 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 y 0.05 0.05 0.05 nsit

0.6 0.6 0.6 te u) in

(a 0.00 0.00 0.00 ze

Si 0.4 0.4 0.4 –0.05 –0.05 –0.05 Normalised 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 Gaia Collaboration et al. 2018, A&A, 616, A1 3.9 ± 0.6 M⊙. for the first time, to observationally constrain the physical properties of the Lindegren, L. et al. 2016, A&A, 595, A4 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 young stars. This new technique of gas content of the disc (Muzerolle et al., ­Arizona Press), 339 Thi, W. F. et al. 2005, A&A, 430, L61 ­combining the spectrum fit with inter­ 2004). ESO/P. Carrillo ESO/P.

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. Davies1 ­circumsecondary discs. During GRAVITY expected location of the dust sublimation Edward Hone1 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 1 Astrophysics Group, University of ponent (CO Ori A) was sufficiently bright ­peri­odicity in its optical photometry ­Exeter, UK (K = 6.0 magnitudes) for standard single ­(Rostopchina et al., 2007), potentially 2 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, Lazareff­ 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 R   A angular and high spectral resolution % using the best-fit model L@ across the K-band allows us to study the   from our modelling SSN  of the continuum visibili- dynamic inner regions of protoplanetary @S HN M ties. Each pixel in the Q KHLH discs and to directly measure the pro-   map is coloured to cesses by which mass is accreted onto l reflect the maximum TOOD  6 #DBKHM possible flux contribu-

stars and launched into outflows. m   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 ­companions 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 )

0.02 ) as 50 –1 0 s e) (m m re on (k

eg –5 0

ati 0.00

(d 5 PA = 54.2 BL = 37.1 PA = 70.85 BL = 120.0 PA = 84.45 BL = 88.8 loci ty DP –50 0 Ve Δ Declin –0.02 –100 –5 2.164 2.1662.168 2.164 2.166 2.1682.164 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 field component strength of ~ 2 kG for has been observed previously by, for cation team in the execution and reduction of our 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 calculate model-independent photocen- GRAVITY’s impact and outlook for the Hill, C. A. et al. 2019, MNRAS, 484, 5810 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 ESO/S.

The landscape of the Chajnantor Plateau on which the antennas of the Atacama Large ­Milli­meter/submillimeter Array (ALMA) are located.

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 Dong1 ­Einstein reluctantly published the idea of 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 1 Kavli Institute for Astronomy and Astro- 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- 2 ESO ideal case of perfect alignment, the bent marily the Galactic bulge). During these 3 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 4 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 5 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 6 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 ° F /F = 0.20 ± 0.02 2 .9 lens source same technique could be applied in the u = 0.268 ± 0.03 – 4 A = 2.42 future to measure the mass and dis- + χ 2 =1.20 = r tance of completely dark objects, such E PA

a black hole. In fact, while black holes ) 1 in binaries have been found from X-ray as

and LIGO gravitational-wave observa- (m on tions, and are likely to be found in the θ 0 S ati future by Gaia astrometry, gravitational u θ E = 0.51 mas microlensing is the only known way to ξ = 0.20 θ

Declin E find isolated black holes. Our detection = Figure 1. Model of the apparent

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

θ apparent sizes of the images, but –2 A =1.42 rather an indication of their respective – 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 10–1 –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 niques, all known stellar-mass BHs are the Spitzer light curves, and by combin- Fukui, A. et al. 2019, AJ, 158, 206 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 arXiv:1912.00038 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 Interferometry

GRAVITY Collaboration (see page 20)

2000 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 as 0

reach a new domain of astrophysics: (m 20 0 exoplanetology. In March 2019, the ion 0 GRAVITY collaboration published the first spectrum and astrometry of an 0

exoplanet obtained by optical interfer- Declin at –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 2000 1000 0 New opportunities, new challenges 20 Right ascension (mas –) 0 –2 With the 2019 Nobel Prize, jointly awarded 10 0 fibre @ 70 mas to Michel Mayor and Didier Queloz, the

s) 20 fibre @ 90 mas

field of exoplanet research received te –0 fibre @ 120 mas worldwide recognition. It is true that for 10 –3 40 fibre @ 150 mas the first 20 years, the domain was more minu 0 fibre @ 200 mas 5 fibre @ 250 mas

akin to a giant search for Easter eggs. /1 The rarity of each discovery meant it had σ (5 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 rast ratio (Transiting Exoplanet Survey Satellite), 10 –5 resulting in the discovery of thousands of Cont exoplanets, and the development of a large community including many young 050 100150 200 250300 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 radial velocity instruments — in which Near-infrared and optical Échelle Spec- AU Mic (data from the InfraRed Dual-band Imager 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 ­SPectrograph for Rocky Exoplanet and line (Alonso-Floriano et al., 2019). How- ­correspond to the field of view of GRAVITY. The 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 planets, revealing gaps such as the hot the transit. In the long term, the most range is extracted and is plotted on the contrast 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 c 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 d GRAVITY as a planet hunter

0.03 GRAVITY’s high dynamic range at Luhman 16A Exo-REM T = 1150 K / log(g) = 4.3 GPI data

­angular separations as small as 100 mas ) is obtained thanks to this exquisite post- –1 0.02 SPHERE photmetry

µm GRAVITY data

processing. Figure 1 shows ­GRAVITY –2 Wm observations of the star AU Mic. The disc 4

–1 0.01 0

of AU Mic has prominent structures, which (1

λ are resolved with SPHERE (Boccaletti, F Thalmann & Lagrange, 2015). GRAVITY 0.00 looks for point-like sources, of size 1.9 2.02.1 2.2 2.32.4 2.5 smaller than its interferometric resolution Wavelength (µm) (< 3 mas). Larger objects are not seen in Figure 2. Upper panel: SPHERE/IRDIS image of the coherent flux of the interferometer. 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 scope (60 mas in the K-band for the VLT K-band spectrum of HR8799e at spectral resolution 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- trograph) is placed at different positions et al., 2019). 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 stellar­ 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 ­fitting a grid of models. The aim is to Acknowledgements Gauchet, L. et al. 2016, A&A, 595A, 31G obtain direct estimates of atomic ratios. GRAVITY Collaboration et al. 2019, A&A, 623, 11 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 (Öberg, Murray-Clay & Bergin, 2011). Wertz, O. et al. 2017, A&A, 598, A83 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 Acknowledgement: 2. Davide de Martin

Wide-field image showing the field in the constellation of Pegasus centred on HR8799.

The Messenger 178 – Quarter 4 | 2019 49 ESO/M. Zamani ESO/M. Astronomical News

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 Christensen1 Figure 2. Alpenglow Petr Horálek1 )ESO seen from above ESO’s Paranal Observatory. The VLT and VISTA domes are coloured red 1 ESO during the sunset. atacamaphoto.com

Several interesting atmospheric phe- nomena take place during dusk and ( G.Hüdepohl 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/ESOR.

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 over Paranal Observa- tory, painting an array of subtle hues across the

ESO/R. Wesson ESO/R. 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, When the Sun sets behind high moun- a partial lunar eclipse was visible in the sky above 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 perfect conditions, when crepuscular rays can be followed all the way across the sky, they appear to expand at first Horálek ESO/P. 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 crescent Moon the solar or lunar disc appears layered. deformed by the atmos- phere over the Pacific The subtle details of the layered lantern P. Horálek/ESO P. 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 atmosphere have different temperatures red (Young, 2013). Where they overlap, stars in this multi-exposure photo: atmospheric ­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/ESOR.

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 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). Horálek/ESO P.

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 such as those seen in Figures 8 and 9, 16 November 2017 shows the phenomena of red seen from Cerro Paranal. The image was taken by (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 image. The ozone layer is responsible for the spectral effect, since the light path through the atmosphere is longer at ESO/S. Guisard ­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 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 Blanchard/ESO G. 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 whole mystery surrounding the flash caused by sunlight backscattering off tiny and makes it something quite special to ESO/C. Malin ESO/C. droplets of water in look for. the atmosphere. Glories appear at a point The green flash is best seen on a com- directly opposite the 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 D.

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 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 Horálek ESO/P. 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 2018): https://www.flickr.com/photos/ bob_81667/39604010580/ 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 sdsu.edu/explain/simulations/std/rims.html 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 rather complicated case of Mie scattering The authors are grateful for helpful conversations Kundt, W. 2001, Current Science, 81, No. 4, 399 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, shadows, they will each see only one Vol. 20, No. 12, 2043 Lock, J. A. 2015, Applied Optics, Vol. 54, Issue 4, B54 Notes glory and claim it to be around their own Longo, G. 2007, Comet/Asteroid Impact and Human head. Looking down at a plane’s shadow Society, 303 a when flying, you will often be able to see Please don’t forget that looking at the Sun itself, Lynch, L. K. & Livingston, W. 2001, Color and Light in especially through an optical device (camera, tele- a glory on the cloud tops. Nature, (Cambridge: Cambridge University Press) 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 doing. 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 ­William Heinemann) observatories makes dawn and twilight de Veer, G. 1876, The Three Voyages of William 1  colours very intense, but also followed by The colors of sunset and twilight (Corfidi, S. F. Barents to the Arctic Regions: 1594, 1595 and 2014, NOAA/NWS SPC): https://www.spc.noaa. a steep darkening gradient. For these 1596, (London: Forgotten Books, 2017) gov/publications/corfidi/sunset/ Waythomas, C. W. et al. 2010, Journal of 2  reasons, very bright objects, such as 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 www.hko.gov.hk/m/article_e.htm?title=ele_00493 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 Christopher Harrison1 the nationalities of the applicants for the first

1 ts Anita Zanella ESO Summer Research 1 Claudia Agliozzo Programme. Richard I. Anderson1 2

Fabrizio Arrigoni Battaia applican 1 50

Francesco Belfiore of Remco van der Burg1 Chian-Chou Chen (T. C.) 1 1 Stefano Facchini umber Jérémy Fensch1 N Prashin Jethwa1 0 s h h h h h h n n Rosita Kokotanekova1 h is tc lish ec rian

1 Iris

Federico Lelli ench edis Italia n Swis Du Po Brit 1 Cz Fr Danish Finnis uguese Belgia Anna Miotello Chilea Spanis German Aust Sw Anna Pala1 rt

Miguel Querejeta1 Po Adam Rubin1 Nationality Dominika Wylezalek1 Laura Watkins1 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, 1 ESO Research Programme at ESO for up to open to all ESO staff, on 1 July 2019. 2 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 website1, 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: Remco 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, amongst these different scenarios it is ­Matthew Wilkinson, ­presents 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 ag 15.8 six observing epochs (m

r between February and

H May 2019. Modulated variability: a new window 16.0 into stellar pulsations Advisors: Richard I. Anderson Student: Samuel Ward (UK), University of 16.2 Durham, UK 0.0 0.20.4 0.6 0.81.0 1.21.4 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: Chian-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: Prashin 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 ­talented 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 dark matter content. 012345results from the feed- 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: Stefano 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- of discs and is a key ingredient in any tive assistants. In particular, we thank Stella-Maria Chasiotis-Klingner for her excellent support with­ 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- 2 to different predictions about the evolu- sented in the ESOBlog 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. 1 ESO summer research programme website: In this project, Francesco tested the vis- http://eso.org/summerresearch/ 2 ESOBlog entry Meet Our 2019 Summer Research cous scenarios by comparing statistical The great success of the programme Programme Students: https://www.eso.org/public/ 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 ESO/S.

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. Boffin1 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 1 ESO ensure that “the machine is being taught 2 Helmholtz-Institut für Strahlen- und to take into account certain relevant Kernphysik, Universität Bonn, Germany known facts” (Griffin 2014). Moreover, the 3 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 ­clustering 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, researchers, and 37% tenure-track for ensuring the best conditions during the meeting. 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- world and approximately two dozen tion about the authors, including their Adorf, H.-M. 1991, The Messenger, 63, 69 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 Notes talks and 50% of the posters being given The Scientific Organising Committee by students. a Machine learning is one of the most commonly worked hard to ensure fair representa- used subsets of AI. tion from the community. Among the Acknowledgements 10 invited speakers, five were female. Links Three of the five sessions were also It is a great pleasure to thank the members of the 1 Workshop website: https://www.eso.org/sci/ chaired by women. Among the abstracts Scientific Organising Committee (Coryn Bailer- meetings/2019/AIA2019.html submitted, a quarter were by women, Jones, Henri Boffin, Massimo Brescia, Torsten­Enßlin, 2 The Kaggle platform website: kaggle.com Emille Ishida, Zdenka Kuncic, Antoine Mérand, and this was also the female/male ratio 3 Numerical Information Field Theory: http://ift. Melissa Ness, and Felix Stoehr) for setting up an among the contributed speakers. We pages.mpcdf.de/nifty/ amazing programme, the invited speakers for 4 2019 AIA workshop programme: https://www.eso. had a very high level of participation from remarkable and clear reviews, and the organisers of org/sci/meetings/2019/AIA2019/Booklet_final.pdf 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, ­registration fee for students and the fact Philipp Frank, Sebastian Hutschenreuter, and ­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 Johnston1 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. 1 ESO discuss their various needs in order to The literature is full of suggestions and 2 Deakin University, Burwood, Australia effectively bridge the gap between advice about how to best communicate 3 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 Salimpour­ 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 working with real data. The lack of user- deaf children and school-age youth in ­balconies inside the ESO Supernova, overlooking 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”

40% 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 Japan

Figure 2. Pie chart showing the distribution of Figure 3. Multi-level pie chart showing the gender ­countries 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 Conference webpage: https://iau-dc-c1.org/ – two 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: https://astroedu.iau.org/en/ 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 questions.

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, I continued to follow my passion for 1 ESO Catch a Star contest: http://www.eso.org/ 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-­type 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 Erratum

P (selfreported | DeepThought) 2DMHNQHSX./" 0.75 e

2DMHNQHSX4RDQR t

  rs

ledg 0.00470.22 0.78

Wo 0.60 ow kn   0.45

Q@BSHNM 0.1 0.35 0.55 inferred %

  t Median 0.30

  Though 0.52 0.37 0.11 0.15 Best Deep   Expert General No knowledge XDS @QR @QR D@QR G# XD X knowledge N/ G@MXD l M - RRS DDM DSG@ +D !DSV ,NQ Referee perceived knowledge

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