Abstract

One distinctive feature of low-luminosity active galactic nuclei (LLAGNs) is the relatively weak reflection features they may display in the X-ray spectrum, which can result from the disappearance of the torus with decreasing accretion rates. Some material, however, must surround the active nucleus, i.e. the accretion flow itself and, possibly, a flattened-out or thinned torus. In this work, we study whether reflection is indeed absent or undetectable due to its intrinsically weak features together with the low statistics inherent to LLAGNs. Here, we focus on NGC3718 (L/L-Edd similar to 10(-5)) combining observations from XMM-Newton and the deepest to date NuSTAR (0.5-79 keV) spectrum of a LLAGNs, to constrain potential reflectors, and analyse how the fitted coronal parameters depend on the reflection model. We test models representing both an accretion disc (Relxill) and a torus-like (MYTorus and Borus) neutral reflector. From a statistical point of view, reflection is not required, but its inclusion allows to place strong constraints on the geometry and physical features of the surroundings: both neutral reflectors (torus) tested should be Compton thin (N-H < 10(23.2)cm(-2)) and preferentially cover a large fraction of the sky. If the reflected light instead arises from an ionized reflector, a highly ionized case is preferred. These models produce an intrinsic power-law spectral index in the range [1.81-1.87], where the torus models result in steeper slopes. The cut-off energy of the power-law emission also changes with the inclusion of reflection models, resulting in constrained values for the disc reflectors and unconstrained values for torus reflectors.