Mid-infrared interferometry with K band fringe-tracking I. The VLTI MIDI+FSU experiment

Context. A turbulent atmosphere causes atmospheric piston variations leading to rapid changes in the optical path difference of an interferometer, which causes correlated flux losses. This leads to decreased sensitivity and accuracy in the correlated flux measurement. Aims. To stabilize the N band interferometric signal in MIDI (MID-infrared Interferometric instrument), we use an external fringe tracker working in K band, the so-called FSU-A (fringe sensor unit) of the PRIMA (Phase-Referenced Imaging and Micro-arcsecond Astrometry) facility at VLTI. We present measurements obtained using the newly commissioned and publicly offered MIDI+FSU-A mode. A first characterization of the fringe-tracking performance and resulting gains in the N band are presented. In addition, we demonstrate the possibility of using the FSU-A to measure visibilities in the K band. Methods. We analyzed FSU-A fringe track data of 43 individual observations covering different baselines and object K band magnitudes with respect to the fringe-tracking performance. The N band group delay and phase delay values could be predicted by computing the relative change in the differential water vapor column density from FSU-A data. Visibility measurements in the K band were carried out using a scanning mode of the FSU-A. Results. Using the FSU-A K band group delay and phase delay measurements, we were able to predict the corresponding N band values with high accuracy with residuals of less than 1 μm. This allows the coherent integration of the MIDI fringes of faint or resolved N band targets, respectively. With that method we could decrease the detection limit of correlated fluxes of MIDI down to 0.5 Jy (vs. 5 Jy without FSU-A) and 0.05 Jy (vs. 0.2 Jy without FSU-A) using the ATs and UTs, respectively. The K band visibilities could be measured with a precision down to ≈2%.