The first two decades of your professional career you worked on the zodiacal light. How did this come about?
By accident. The zodiacal light is a measure of interplanetary dust. Normally, for these studies, there was no need for better than about a degree of spatial resolution, so cm-sized optics was sufficient. This sounds very simple, but zodiacal light observations require particular care in calibration and detailed foreground and background subtraction. Reliable brightness and polarisation data became available in the early 1960s. At that time, for a physics student with the goal of pursuing astronomy, zodiacal light would not appear as an attractive start of a career. However, the situation changed when in 1965 German chancellor Ludwig Erhard visited US president Lyndon B. Johnson and brought back an agreement of a German-US solar probe project, a very large step forward for space science in Germany. It would provide PhD students and postdocs, like myself, the fortunate opportunity to contribute - even as team leaders of experiments - to the up-to-date interplanetary space probes called Helios 1 and Helios 2. They were launched in 1974 and 1976 and approached the Sun to within 0.29 AU. The zodiacal light instrument on these probes, which was taken over by the Max-Planck-Institut für Astronomie in Heidelberg, sounded like a rewarding topic for my thesis.
What triggered your interest in young stars?
Actually, my interest in young stars developed gradually, when around 1980 my activities on the zodiacal light declined. At that time Mel Dyck from the Institute for Astronomy in Hawaii worked for two years at our Max-Planck-Institute. Shortly before he arrived, we got the surprising message that at the end of 1981 he had discovered, by means of slit-scan speckle interferometry in the near-infrared, a companion 0.6'' south of T Tauri. I joined Mel Dyck in building a dedicated transportable slit-scanning near-infrared speckle interferometer giving spatial resolutions of 0.1'' to 0.2'' on the 2.2 m and 3.6 m telescopes available to us on Calar Alto in Spain and La Silla in Chile. Obviously, one of the object groups to be studied in a similar way like T Tau were the low-mass young T Tauri stars. In the early 1980s, important elements were added to our understanding of T Tauri stars: obscuring circumstellar disks were quite common among them and some showed narrow high-velocity jets, for example, -250 km/s in the case of DG Tau. For us, this pointed to the value of studying the vicinity of young, low-mass stars. And the technique of lunar occultations, brought to us by Andrea Richichi from Firenze, also proved very useful for our studies. The speckle equipment for scan length zero provided the needed fast photometry. In addition, Michal Simon from Stony Brook spent a sabbatical year at our institute in 1991, starting a lasting exchange of discussion on the physical meaning of our work. Needless to mention that the continued cooperation with Hans Zinnecker, then at the University of Würzburg, the tireless defender of the importance of binarity, pushed us to emphasize in our studies the duplicity of young stars.
Which were the first results you particularly remember from your studies?
These were contributions to the study of “Infrared Companions”. The term was already used in the title of the paper announcing the detection of the T Tauri companion, which could not be seen in the visual but surpassed the brightness of the primary in the infrared beyond 3-4 μm. Such changes of relative brightness were also true for the infrared companions to the T Tauri stars DoAr 24E and Glass-I found by Zinnecker and Chelli and for Haro 6-10 with its companion found by Martin Haas and myself as part of our programme in 1988. At that time these objects, presumed of the same age as their visual “primaries”, promised the exciting possibility of a new evolutionary path for young stars, in which more massive components could evolve more slowly, opposite to the expected behaviour. By 1997, however, as summarised by Chris Koresko from McDonald Observatory with Tom Herbst and myself, the view had generally emerged that the infrared companions could represent relatively normal young low-mass stars, where strong and often variable local extinction and accretion could explain the apparently unusual phenomenon.
In 1993 you published an enormously influential paper on statistics of young binaries in Taurus, one of three papers on PMS binaries that appeared from different groups within one month of each other.
Preliminary versions of these three papers had been given at the IAU colloquium 135 in April 1992 in Georgia, USA with Harold McAlister and William Hartkopf as editors of the proceedings. The two papers by Andrea Ghez on T Tauri stars in Taurus and Ophiuchus and by myself on young stars in Taurus were at 2.2 μm. The third one by Hans Zinnecker and yourself, mainly on pre-main sequence stars in nearby star-forming regions in the southern hemisphere, was in the I band at 0.95 μm. To this Michal Simon from Stony Brook added a lunar occultation and imaging survey in Taurus and Ophiuchus, and Bob Mathieu a radial velocity survey of young binary systems with “close” separation, i.e. with periods of < 100 days. So, if one had to define when exactly the importance of duplicity in low-mass star formation became generally accepted, my choice would be the opening session of this IAU colloquium. In those days, several developments contributed to the quick progress of the field. Considerable improvement of near-infrared detectors helped to perform such surveys with meaningful area coverage and sensitivity. New techniques like speckle interferometry and later adaptive optics allowed reaching the diffraction limit of the telescopes. But also the multiplicity of G-type main-sequence stars in the solar neighbourhood had been reliably determined by Antoine Duquennoy and Michel Mayor with the radial velocity instrument CORAVEL in 1991. This gave a solid reference to which one could compare the young binaries. Also, the simplicity of the results was important for their acceptance. For example, the Heidelberg sample of 104 T Tauri stars led to a duplicity in Taurus of 42% +/- 6%, higher by a factor of 1.9 in the separation range of 0.13'' - 10'' (18-1400 AU) than found in Duquennoy and Mayor's G-dwarf sample for these separations. Assuming for both studies the same binary separation distribution, this corresponds within the error bars exactly to the result that all young low-mass stars (101% +/- 15%) would be born double. In addition, the distribution of companions with separation indeed was similar in the two samples. Nor was there a systematic difference in duplicity between “classical” T Tauri stars and weak-lined T Tauri stars. This was clearly seen when Rainer Köhler added 76 WTTSs detected by the X-ray satellite ROSAT to the Heidelberg Taurus sample. The original view, that all star-forming regions might show the same high duplicities of their low-mass stars, however, did not hold. As indicated in a common paper by Simon, Ghez, and myself in 1995, the duplicity in Ophiuchus barely showed an excess to the G dwarf sample, just by a factor of 1.1 +/- 0.3. With a sample of 158 young low-mass stars in Ophiuchus, Thorsten Ratzka confirmed this trend of lower multiplicity for denser star-forming regions. Trends with the mass of the primary or with the mass ratio in young binary or multiple systems also provide tests on star formation. It all came together at the IAU symposium 200 in Potsdam in April 2000 on the formation of binary stars with the two editors Hans Zinnecker and Bob Mathieu. And the Annual Reviews article by Gaspard Duchêne and Adam Kraus on stellar multiplicity in 2013 show that young low-mass binaries remain an important topic in star formation.
In the following years, you carried out a study of the binarity and near-infrared halos of Herbig AeBe stars. Did you find a difference to the T Tauri stars?
Herbig Ae/Be stars and similar objects have an advantage for studies at high spatial resolution because of their higher luminosity and brightness with respect to T Tauri stars. We started looking at individual sources with interesting structures, mainly halos. Step by step we obtained a sample of 31 objects. The incidence of duplicity among them was similar to the one for T Tauri stars in Taurus, but the existence of near-infrared halos was much more common. These halos were apparently due to scattering of stellar near-infrared radiation from the inner parts of circumstellar material. However, the detection of 1.3 mm continuum emission from circumstellar dust by Steven Beckwith and Anneila Sargent in 1990 was mostly around T Tauri stars.
In 2004 you published another paper on disks around Herbig AeBe stars, this time as observed at mid-infrared wavelengths. What did you conclude?
In the 1990s, model fits of the spectral energy distribution, particularly in the infrared, resulted in increasingly detailed predictions for the spatial distribution of circumstellar dust around Herbig Ae/Be stars. The usual problem is the ambiguity in temperature and density distribution. The model of a group in Amsterdam, led by Rens Waters, had as an important feature that the innermost ring of a circumstellar disk gets “puffed up” by the high irradiation from the star. Thus it would be casting a shadow on the outer regions. Objects with flaring disks, called group I, therefore would look redder and larger at mid-infrared wavelengths than objects without, called group II. This geometry should allow reducing the spatial ambiguity. We checked the predicted colour-size correlation using the 10 micron mid-infrared interferometer MIDI on the ESO Very Large Telescope. It was qualitatively confirmed during the early observations of a small sample of seven stars. The observed half-light radius was growing from approximately 1 AU to above 5 AU with increasing IRAS colour (flux(25 μm)/flux(12 μm)).
Together with Uwe Graser you formed the PI team for the development of the mid-infrared interferometric instrument MIDI for the VLTI at Paranal. What were the capabilities of this instrument and what are some of the subsequent results on young stars that you have been involved with?
From its beginning, the VLT was planned as a telescope system composed of four 8-m dishes, with interferometric measurements being a standard part of the operation. In 1997, when the interferometric instruments got defined, our group at MPI-A, as leader of a consortium with Dutch, French, and German contributing institutes, was selected to develop the mid-infrared interferometer MIDI (8 μm - 13.5 μm). Earlier attempts in this wavelength range had failed because of lack of stability in the presence of the overwhelming background radiation from sky, telescope, and the instrument. In MIDI, the latter resulted from 23 reflecting or absorbing optical elements along the light path from the main mirror to the detector. The angular resolution in the mid-infrared on the 130 meter baseline between the VLT telescopes UT1 and UT4 at a wavelength of 10 μm was 15 milliarcseconds. The sensitivity resulting from the large single-telescope mirror size allowed observing sources down to 0.5 Jy, corresponding to N=5 mag, the brightness of the well-studied T Tauri star TW Hya. Here, the mid-infrared interferometry showed that the optically thick circumstellar disk extended closer in to the star than inferred at millimeter wavelengths. The MIDI instrument also offered spectral capabilities with two low to moderately low resolutions of λ/Δλ ≈30 and ≈130. Of these, because of the higher sensitivity, the first one was the usual choice and the standard observing mode with MIDI. A central aspect of the MIDI project - when compared to earlier similar-sized projects - was the enormous increase in the importance and the capabilities of the software. Just to mention the ingenious MIDI data reduction system called EWS conceived and built by Walter Jaffe from the project partner Sterrewacht Leiden. It measures visibilities, not their square, and it allows them to determine relative phases along the spectrum. Thereby the linear part of the phase spectrum - corresponding to the position of the source in the field - is set to zero. Also, with the information available on the VLTI, the before/after uncertainty in the scan can be broken. In regard to observations of circumstellar disks of young stars, MIDI directly gave information on the existence and size of silicate emission or absorption regions. In addition, crystalline olivine - at 11.3 μm - and crystalline pyroxene - at 9.2 μm - also can be detected by their features in the N band. To give again an example of observations of a few Herbig Ae stars during early observations with MIDI: HD 144432, HD 163294 and HD 142527 were known to harbour crystalline silicates in their disks. The collaborating Amsterdam group mentioned earlier showed in 2004 that the dust grains in the inner 2 AU of these circumstellar disks are much more crystallized than in the outer parts, almost completely so in the case of HD 142527, and mostly due to olivine. When the MIDI operations - 2002 to 2016 - came close to their nominal end, summaries of larger samples were possible. For example, Jonathan Menu and Roy van Boekel from our group at MPI-A concluded in 2015 that our sample of 38 Herbig Ae/Be stars indicated a possible evolution from group II (with continuous disk) into group I sources through an intermediate group II with gaps in the disk, as it might result from interaction with a massive planet. MIDI on the VLTI is now being followed by the second generation mid-infrared interferometer MATISSE with Principal Investigator Bruno Lopez from the Observatoire de la Côte d'Azur. The potential of the imaging capabilities presented by MATISSE will soon be seen, including studies of luminous young stars.