Two Populations of X-ray Pulsars Produced by Two Types of Supernovae

Two types of supernova are thought to produce the overwhelming majority of neutron stars in the Universe. The first type, iron-core collapse supernovae, occurs when a high-mass star develops a degenerate iron core that exceeds the Chandrasekhar limit. The second type, electron-capture supernovae, is associated with the collapse of a lower-mass oxygen-neon-magnesium core as it loses pressure support owing to the sudden capture of electrons by neon and/or magnesium nuclei. It has hitherto been impossible to identify the two distinct families of neutron stars produced in these formation channels. Here we report that a large, well-known class of neutron-star-hosting X-ray pulsars is actually composed of two distinct sub-populations with different characteristic spin periods, orbital periods and orbital eccentricities. This class, the Be/X-ray binaries, contains neutron stars that accrete material from a more massive companion star. The two sub-populations are most probably associated with the two distinct types of neutron-star-forming supernovae, with electron-capture supernovae preferentially producing system with short spin period, short orbital periods and low eccentricity. Intriguingly, the split between the two sub-populations is clearest in the distribution of the logarithm of spin period, a result that had not been predicted and which still remains to be explained.

💡 Research Summary

The authors investigate a well‑studied class of high‑mass X‑ray binaries known as Be/X‑ray binaries (BeXs), in which a neutron star accretes matter from a rapidly rotating, massive Be‑type companion. By compiling a comprehensive catalogue of spin periods (P_spin), orbital periods (P_orb), and, where available, orbital eccentricities (e), they perform a statistical analysis that reveals a clear bimodal distribution in the logarithm of the spin period. Using the KMM algorithm and non‑parametric tests, they demonstrate that the spin‑period histogram is best described by two Gaussian components centred at roughly 10 s and 200 s, with comparable dispersions (~0.4 dex). The short‑spin group accounts for about 35 % of the sample, while the long‑spin group makes up the remaining 65 %.

The bimodality is also hinted at in the orbital‑period distribution, but it is far less pronounced than in spin. The authors argue that this cannot be explained by two distinct equilibrium spin states that are accessible at any orbital period, nor by evolutionary changes in orbital period during the BeX phase (the timescale for wind‑driven orbital evolution is orders of magnitude longer than the BeX lifetime). Consequently, they propose that the two sub‑populations correspond to two distinct formation channels for the neutron stars.

The two channels are identified with the two principal supernova mechanisms that produce most neutron stars: (1) iron‑core collapse supernovae (CCSNe) from massive stars (> ~10 M☉) and (2) electron‑capture supernovae (ECSNe) from intermediate‑mass progenitors (≈ 8–10 M☉) that develop O‑Ne‑Mg cores. ECSNe are expected to generate relatively low‑mass neutron stars (< 1.3 M☉) and to impart small natal kicks (< 50 km s⁻¹), whereas CCSNe produce slightly heavier neutron stars (~1.4 M☉) and large kicks (> 200 km s⁻¹). In binary systems, the magnitude of the kick directly influences the post‑supernova orbital eccentricity. Hence, ECSN‑origin BeXs should have low eccentricities and short spin periods (because the neutron star settles into a shorter equilibrium spin in a tighter, less disturbed orbit), while CCSN‑origin BeXs should display higher eccentricities and longer spin periods.

The authors test this hypothesis by examining the limited set of systems with measured eccentricities (≈ 20 objects). They find that the long‑spin group tends to have higher eccentricities than the short‑spin group, and a Kolmogorov–Smirnov test yields marginal significance, consistent with the proposed scenario. Moreover, the relative fractions of the two groups are similar in the Small Magellanic Cloud (SMC) and the Milky Way/Large Magellanic Cloud (MW+LMC), arguing against selection effects that might preferentially hide intermediate‑period systems.

The paper concludes with several concrete predictions: (i) short‑spin BeXs (ECSN channel) should have systematically lower space velocities than long‑spin BeXs; (ii) neutron‑star masses in short‑spin systems should be measurably lower; (iii) expanding the sample of systems with reliable P_spin, P_orb, and e measurements will sharpen the statistical case; and (iv) detailed binary‑population synthesis models should be refined to reproduce the observed bimodality and the comparable abundances of the two groups across different galaxies.

Overall, the study provides the first observational evidence that the Be/X‑ray binary population retains a fossil record of the two dominant supernova pathways, and it establishes spin period as a surprisingly sensitive diagnostic of neutron‑star formation channel. This work bridges high‑energy astrophysics, massive‑star evolution, and supernova theory, opening new avenues for testing models of natal kicks, neutron‑star mass distribution, and binary evolution through future X‑ray, optical, and radio observations.