The nightmare scenario: measuring the stochastic gravitational wave background from stalling massive black hole binaries with pulsar timing arrays

Massive black hole binaries, formed when galaxies merge, are among the primary sources of gravitational waves targeted by ongoing pulsar timing array (PTA) experiments and the upcoming space-based Laser Interferometer Space Antenna (LISA) interferometer. However, their formation and merger rates are still highly uncertain. Recent upper limits on the stochastic gravitational wave background obtained by PTAs are starting to be in marginal tension with theoretical models for the pairing and orbital evolution of these systems. This tension can be resolved by assuming that these binaries are more eccentric or interact more strongly with the environment (gas and stars) than expected, or by accounting for possible selection biases in the construction of the theoretical models. However, another (pessimistic) possibility is that these binaries do not merge at all, but stall at large (similar to pc) separations. We explore this extreme scenario by using a semi-analytic galaxy formation model including massive black holes (isolated and in binaries), and show that future generations of PTAs will detect the stochastic gravitational wave background from the massive black hole binary population within 10-15 yr of observations, even in the 'nightmare scenario' in which all binaries stall at the hardening radius. Moreover, we argue that this scenario is too pessimistic, because our model predicts the existence of a subpopulation of binaries with small mass ratios (q less than or similar to 10(-3)) that should merge within a Hubble time simply as a result of gravitational wave emission. This subpopulation will be observable with large signal-to-noise ratios by future PTAs thanks to next-generation radio telescopes such as Square Kilometre Array or Five-hundred-meter Aperture Spherical Telescope, and possibly by LISA.