The Araucaria project: High-precision orbital parallax and masses of eclipsing binaries from infrared interferometry

Abstract

Context. The precise determinations of stellar mass at 1 % provide important constraints on stellar evolution models. Accurate parallax measurements can also serve as independent benchmarks for the next Gaia data release. Aims. We aim at measuring the masses and distance of binary systems with a precision level better than 1 % using a fully geometrical and empirical method. Methods. We obtained the first interferometric observations for the eclipsing systems AI Phe, AL Dor, KW Hya, NN Del, ψ Cen and V4090 Sgr with the VLTI/PIONIER combiner, which we combined with radial velocity measurements to derive their threedimensional orbit, masses, and distance. Results. We determined very precise stellar masses for all systems, ranging from 0.04 % to 3.3 % precision level. We combined these measurements with stellar effective temperature and linear radius to fit stellar isochrones models and determined the age of the systems. We also derived the distance to the systems with a precision level as high as 0.4 %.Conclusions. The comparison of theoretical models with stellar parameters shows that stellar models are still deficient in simultaneously fitting the stellar parameters (T eff , R and M) with such level of precision on individual masses. This stresses the importance of precisely measuring the stellar parameters to better calibrate stellar evolution models. The precision of our model-independent orbital parallaxes varies from 24µas as to 70µas and they provide a unique opportunity to check on the future Gaia measurements for possible systematic errors.

The TESS light curve of AI Phoenicis

“…The HARPS RVs were first corrected for the motion of the binary system due to a third body in the system with a long, eccentric orbit around AI Phe (P 3 100 y, e ∼ 0.8). As Gallenne et al (2019) note the agreement between the parameters derived from these three independent studies is extraordinarily good. We used the following weighted-mean values for the semi-amplitudes to calculate the masses and radii of AI Phe: K 1 = 51.164 ± 0.007 km s −1 and K 2 = 49.106 ± 0.010 km s −1 .…”

“…Sybilski et al (2018) also obtained spectroscopic orbits for both components of AI Phe together with spectroscopic observations obtained during the secondary eclipse that show that the rotation axis of the subgiant component is not aligned with the orbital axis of the binary system. Gallenne et al (2019) have used the VLTI interferometer to measure the astrometric orbit of AI Phe. They used this astrometric orbit combined with their own high-quality spectroscopic orbits for the two stars to derive their masses with a precision of better than 0.1 per cent and a modelindependent distance to the binary of 169.4 ± 0.7 pc.…”

“…At a level of 200 ppm per cadence, which is indicative of what we might expect with PLATO, we would expect to detect high-quality asteroseismic signals from both components of AI Phe. Hełminiak et al (2009), Sybilski et al (2018), and Gallenne et al (2019) have all published spectroscopic orbits for both stars based on RVs measured from spectra with good signal-to-noise ratio (S/N 30) obtained onéchelle spectrographs with a resolving power R 60 000. The spectroscopic orbit by Gallenne et al (2019) was determined from a simultaneous fit to 33 RVs obtained with the HARPS spectrograph together with the astrometric orbit fit to measurements of the relative positions of the two stars obtained with the VLTI interferometer.…”

“…In particular, e and ω are determined from the spectroscopic orbit and light curve independently to good precision. The values of ecos ω and esin ω from the spectroscopic orbits of Sybilski et al (2018) and Gallenne et al (2019) are compared to the values obtained from the light curve in Fig. 6.…”

Fundamental effective temperature measurements for eclipsing binary stars – I. Development of the method and application to AI Phoenicis

“…For the flux ratio in the TESS band we take the mean value with its standard deviation from the results presented in Maxted et al (2020) using a variety of analysis methods. We also include the flux ratio value 2.012 measured in the H-band using the VLTI interferometer by Gallenne et al (2019). The bandpass for this flux ratio measurement is not well defined so we assign an nominal error to this value of 0.01, and use the 2MASS H-band to calculate the flux ratio for the measurement.…”