Direct interferometry in the thermal infrared is planned for both ground based observations, with the 10 micron interferometric mode of the VLT (through its mid-infrared focal instrument, called "MIDI", (Leinert & Graser 1998)), the 10 micron nulling mode on the Keck telescopes (Colavita 1998), and for space based projects mostly dedicated to the characterization of extrasolar planets, such as DARWIN (Leger et al. 1996) or TPF (Angel & Woolf 1997). High angular resolution in the mid-infrared is indeed required in various fields of astrophysics, from the observation and modeling of circumstellar dust shells around late type stars, the study of young stellar objects (Natta 1997), of broad line regions of active galactic nuclei (Voit 1997), to the detection of key spectroscopic features in the atmosphere of extrasolar planets. Yet direct interferometry at wavelengths longer than 2.4 microns has been seldom demonstrated so far, and the problem of fringe visibility calibration in the thermal regime has never been addressed.
In the thermal infrared, ground based interferometric stellar observations face indeed very specific technical difficulties, when compared to the ones encountered in the visible or at near infrared (m) wavelengths. The seeing is much better, and the constraints on the optical surfaces are relaxed, but the interferometric signal is contaminated by an incoherent thermal background, that needs to be minimized, monitored, and properly subtracted in order to achieve high accuracy visibility measurements. Around 10 microns, as discussed in Sect. 3.1, the thermal background signal is several orders of magnitude higher than the stellar flux itself. The difficulties in the alignment of the optics and the lower efficiency of the detectors are also probably responsible for the lack of interferometric observations in the thermal regime. First interferometric stellar observations were obtained by single aperture interferometry (McCarthy & Low 1975; McCarthy et al. 1977). First direct observations with separated apertures followed a decade later with the pioneer experiment of spatio-spectral interferometry conducted by Gay and Mékarnia at CERGA (Gay & Mékarnia 1988; Mékarnia & Gay 1990). But most of the scientific results derived from interferometric observations in the thermal infrared have been obtained through heterodyne interferometry at 11.15 µm (Danchi et al. 1994), studying dust shell properties of bright late type stars. Since only the coherent part of the light is coupled to a local oscillator, this latter technique provides visibility measurements which are relatively insensitive to seeing conditions and then more accurate. Although it has a better adaptability to multi-telescope combination, this technique has a low sensitivity for a small number of telescopes, limited by the narrow usable spectral bandwidth (about 6 GHz, i.e. a spectral resolution of about 5000 at 10 microns). Direct broadband interferometry between large telescopes using adaptive optics and/or spatial filtering for real time correction of atmospheric effects, as planned for instance for the MIDI instrument, should lead to accurate visibility measurements and higher sensitivity. The expected background noise limited magnitude in the N band on two 8m telescopes is about 5, without integration capability, and could in principle reach 12 with a fringe tracker (Leinert & Graser 1998).
Interferometric observations in the L band reported here with the "TISIS" (Thermal Infrared Stellar Interferometric Set-up) experiment face the specific problems of the thermal regime, under the less severe conditions of the 3.4-4.1 µm region. They constitute a first step towards future 10 micron observations and should help define signal processing procedures. More specifically, the aim of these observations is twofold:
Besides, observing in the L or M atmospheric windows has its own scientific interest by looking at structures around 1000 K, such as the inner edges of extended dust shells surrounding late type stars, as in the case of o Ceti or R Leo. Comparing these observations with photospheric measurements in K, and with former or simultaneous 10 µm heterodyne observations would be of great interest in the modeling of complex shells around such objects. Coupling these thermal infrared data with quasi-simultaneous observations in the J,H and K bands obtained with the new IOTA NICMOS camera (Millan-Gabet 1998) would obviously also be very informative.
TISIS largely benefits from the FLUOR (Fiber Linked Unit for Optical Combination) optics (Coude du Foresto et al. 1998), acquisition software, and reduction procedures (Coude du Foresto et al. 1997) and uses the single mode fluoride glass couplers developed for FLUOR (Monerie et al.1985), i.e. optimized for the K band. It also benefits from the whole infrastructure of the IOTA interferometer (Carleton et al. 1994). This enables us to concentrate on the issues specific to thermal infrared interferometry.
In the following sections we present the experimental context, from the IOTA environment to the fiber coupler. We give some instrumental results in relation with the two items above, and finally give visibility measurements on Her, discussing related noise sources.
© European Southern Observatory (ESO) 1999
Online publication: May 6, 1999