Gravitational-wave memory and pulsar timing arrays

Pulsar timing arrays (PTAs) are designed to detect gravitational waves with periods from several months to several years, e.g. those produced by wide supermassive black hole binaries in the centres of distant galaxies. Here, we show that PTAs are also sensitive to mergers of supermassive black holes. While these mergers occur on a time-scale too short to be resolvable by a PTA, they generate a change of metric due to non-linear gravitational-wave memory which persists for the duration of the experiment and could be detected. We develop the theory of the single-source detection by PTAs, and derive the sensitivity of PTAs to the gravitational-wave memory jumps. We show that mergers of 10(8) M(circle dot) black holes are 2 - sigma-detectable (in a direction-, polarization- and time-dependent way) out to comoving distances of similar to 1 billion light-years. Modern prediction for black hole merger rates imply marginal to modest chance of an individual jump detection by currently developed PTAs. The sensitivity is expected to be somewhat higher for futuristic PTA experiments with the Square Kilometre Array.