Estimating the fossil disc mass during supermassive black hole mergers: the importance of torque implementation

In this paper, we revisit the issue of estimating the `fossil' disc mass in the circumprimary disc, during the merger of a supermassive black hole binary. As the binary orbital decay speeds up due to the emission of gravitational waves, the gas in the circumprimary disc might be forced to accrete rapidly and could in principle provide a significant electromagnetic counterpart to the gravitational wave emission. Since the luminosity of such flare is proportional to the gaseous mass in the circumprimary disc, estimating such mass accurately is important. Previous investigations of this issue have produced contradictory results, with some authors estimating super-Eddington flares and large disc mass, while others suggesting that the `fossil' disc mass is very low, even less than a Jupiter mass. Here, we perform simple 1D calculations to show that such very low estimates of the disc mass are an artefact of the specific implementation of the tidal torque in 1D models. In particular, for moderate mass ratios of the binary, the usual formula for the torque used in 1D models significantly overestimates the width of the gap induced by the secondary and this artificially leads to a very small leftover circumprimary disc. Using a modified torque, calibrated to reproduce the correct gap width as estimated by 3D models, leads to fossil disc masses of the order of one solar mass. The rapid accretion of the whole circumprimary disc would produce peak luminosities of the order of 1-20 times the Eddington luminosity. Even if a significant fraction of the gas escapes accretion by flowing out the secondary orbit during the merger (an effect not included in our calculations), we would still predict close to Eddington luminosities that might be easily detected.