Long-term evolution of three-planet systems to the post-main sequence and beyond

We study the stability of systems of three giant planets orbiting 3-8 M☉ stars at orbital distances of >10 au as the host star ages through the main sequence (MS) and well into the white dwarf (WD) stage. Systems are stable on the MS if the planets are separated by more than ~9 Hill radii. Most systems surviving the MS will remain stable until the WD phase, although planets scattered on to small pericentres in unstable systems can be swallowed by the expanding stellar envelope when the star ascends the giant branches. Mass-loss at the end of the asymptotic giant branch triggers delayed instability in many systems, leading to instabilities typically occurring at WD cooling ages of a few 100 Myr. This instability occurs both in systems that survived the star's previous evolution unscathed, and in systems that previously underwent scattering instabilities. The outcome of such instability around WDs is overwhelmingly the ejection of one of the planets from the system, with several times more ejections occurring during the WD phase than during the MS. Furthermore, few planets are scattered close to the WD, just outside the Roche limit, where they can be tidally circularized. Hence, we predict that planets in WD systems rarely dynamically evolve to become `hot Jupiters'. Nor does it appear that the observed frequency of metal pollution in WD atmospheres can be entirely explained by planetesimals being destabilized following instability in systems of multiple giant planets, although further work incorporating low-mass planets and planetesimals is needed.