The expected performance of nulling at the VLTI down to 5 mas

While VLTI offers the recombination of four 8-m telescopes with baselines of more than 100m, it has never hosted a dedicated high-contrast nulling beam-combiner. The SCIFY project aims to design, build and commission Hi-5, the first nulling beam-combiner of the VLTI, optimized for the detection and characterization of young giant exoplanets near the snow line, with spectroscopy up to R=2000 in the L' band. It will make use of advanced four-beam nulling combination schemes, like double-Bracewell and kernel-nulling implemented in a single-mode photonic device to produce differential nulled outputs with self-calibrating properties. In the wavelength range of interest, both instrumental errors and background noise are significant. In order to estimate the practical performance of these different configurations in the presence of instrumental errors and further optimize the instrumental design, we have developed SCIFYsim. SCIFYsim is an end-to-end simulator geared towards single-mode beam combiners with of a wide variety of instrumental errors, like optical path difference residuals from fringe tracking, wavefront error at the injection, longitudinal dispersion, chromaticity of the combiner chip, and more. In order to evaluate the performance of the combined spectral channels, we use statistical tests based on likelihood ratios, and account for the covariance in the data. In this paper, we present the expected performance of Hi-5 with a few examples and discuss the main technical limitations to reach the contrast required to image young giant exoplanets.