2024MNRAS.531.2379S

Core-collapse supernovae (CCSNe) are considered the primary magnetar formation channel, with 15 magnetars associated with supernova remnants (SNRs). A large fraction of these should occur in massive stellar binaries that are disrupted by the explosion, meaning that $\sim 45~{{\ \rm per\ cent}}$ of magnetars should be nearby high-velocity stars. Here, we conduct a multiwavelength search for unbound stars, magnetar binaries, and SNR shells using public optical (uvgrizy bands), infrared (J, H, K, and K_s bands), and radio (888 MHz, 1.4 GHz, and 3 GHz) catalogues. We use Monte Carlo analyses of candidates to estimate the probability of association with a given magnetar based on their proximity, distance, proper motion, and magnitude. In addition to recovering a proposed magnetar binary, a proposed unbound binary, and 13 of 15 magnetar SNRs, we identify two new candidate unbound systems: an OB star from the Gaia catalogue we associate with SGR J1822.3-1606, and an X-ray pulsar we associate with 3XMM J185246.6 + 003317. Using a Markov Chain Monte Carlo simulation that assumes all magnetars descend from CCSNe, we constrain the fraction of magnetars with unbound companions to $5\lesssim f_u \lesssim 24~{{\ \rm per\ cent}}$, which disagrees with neutron star population synthesis results. Alternate formation channels are unlikely to wholly account for the lack of unbound binaries as this would require $31\lesssim f_{nc} \lesssim 66~{{\ \rm per\ cent}}$ of magnetars to descend from such channels. Our results support a high fraction ($48\lesssim f_m \lesssim 86~{{\ \rm per\ cent}}$) of pre-CCSN mergers, which can amplify fossil magnetic fields to preferentially form magnetars.