[Paper Review] The millisecond pulsar mass distribution: Evidence for bimodality and constraints on the maximum neutron star mass

John Antoniadis, Thomas M. Tauris+4 more|arXiv (Cornell University)|May 5, 2016

Pulsars and Gravitational Waves ResearchPhysics and Astronomy15 references102 citations

TL;DR

Using a Bayesian analysis of MSP masses, the authors find a strongly bimodal mass distribution and derive a lower limit on the maximum neutron star mass (M_max ≥ 2.018 M⊙ at 98% CL), with implications for the neutron star equation of state and future surveys.

ABSTRACT

The mass function of neutron stars (NSs) contains information about the late evolution of massive stars, the supernova explosion mechanism, and the equation-of-state of cold, nuclear matter beyond the nuclear saturation density. A number of recent NS mass measurements in binary millisecond pulsar (MSP) systems increase the fraction of massive NSs (with $M > 1.8$ M$_{\odot}$) to $\sim 20\% $ of the observed population. In light of these results, we employ a Bayesian framework to revisit the MSP mass distribution. We find that a single Gaussian model does not sufficiently describe the observed population. We test alternative empirical models and infer that the MSP mass distribution is strongly asymmetric. The diversity in spin and orbital properties of high-mass NSs suggests that this is most likely not a result of the recycling process, but rather reflects differences in the NS birth masses. The asymmetry is best accounted for by a bimodal distribution with a low mass component centred at $1.393_{-0.029}^{+0.031}$ M$_{\odot}$ and dispersed by $0.064_{-0.025}^{+0.064}$ M$_{\odot}$, and a high-mass component with a mean of $1.807_{-0.132}^{+0.081}$ and a dispersion of $0.177_{-0.072}^{+0.115}$ M$_{\odot}$. We also establish a lower limit of $M_{max} \ge 2.018$ M$_{\odot}$ at 98% C.L. for the maximum NS mass, from the absence of a high-mass truncation in the observed masses. Using our inferred model, we find that the measurement of 350 MSP masses, expected after the conclusion of pulsar surveys with the Square-Kilometre Array, can result in a precise localization of a maximum mass up to 2.15 M$_{\odot}$, with a 5% accuracy. Finally, we identify possible massive NSs within the known pulsar population and discuss birth masses of MSPs.

Motivation & Objective

• Assess the intrinsic mass distribution of millisecond pulsars (MSPs) from updated mass measurements.
• Test single vs. multi-component models to determine the best representation of MSP masses.
• Infer implications for the neutron star equation of state (EoS) and maximum mass constraints.
• Forecast how future surveys (e.g., SKA) could refine M_max estimates.

Proposed method

• Compute MSP mass likelihoods from heterogeneous measurements (precise masses, total-mass constraints, and mass ratios).
• Compare empirical mass distribution models (single Gaussian vs. bimodal Gaussian) using Bayesian inference and MCMC sampling.
• Use AICc for model selection to penalize model complexity.
• Incorporate a truncation mass M_max to constrain the high-mass end.

Experimental results

Research questions

• RQ1 Is the MSP mass distribution better described by a single Gaussian or a bimodal (two-Gaussian) model?
• RQ2 What are the parameters (means, dispersions, relative contributions) of the preferred MSP mass distribution?
• RQ3 Does the MSP mass distribution permit a meaningful lower limit on the maximum neutron star mass M_max, given current data?
• RQ4 What are the implications of the inferred distribution for the neutron star equation of state, and how might future surveys tighten M_max constraints?

Key findings

• The data favor a bimodal MSP mass distribution over a single Gaussian, as quantified by model comparison (AICc favors Model II).
• The best-fitting bimodal model has components centered at μ1 = 1.396 M⊙ with σ1 = 0.045 M⊙ and μ2 = 1.807 M⊙ with σ2 = 0.177 M⊙, with relative weight r = 0.425.
• The separation statistic D indicates a likely distinct high-mass component, with D = 3.12^{+1.95}_{-1.61} and 73% of samples having D ≥ 2.
• There is a robust lower limit on the maximum NS mass: M_max ≥ 2.018 M⊙ at 98% confidence level (and ≥ 1.924 M⊙ at 99.98% CL).
• The high-mass tail and absence of a high-mass truncation in the current MSP sample underpin the M_max constraint; future SKA-era samples (~350 MSPs) could localize M_max to ~2.15 M⊙ with ~5% accuracy.
• The analysis considers alternative birth-accretion scenarios (Model III) but finds them less favored than the bimodal interpretation for explaining the observed MSP masses.