Identification and characterization of optical companions to the population of millisecond pulsars in the globular cluster M3

TL;DR

The paper develops and applies a joint radio-optical approach to identify and characterize the companions of five binary millisecond pulsars in the globular cluster M3. By leveraging precise phase-connected timing from FAST (Li et al. 2024) and deep, multi-band HST imaging, the authors perform accurate astrometric cross-matching and place the counterparts on helium WD cooling tracks, using MCMC-based modeling to infer companion and neutron star properties. They confirm M3B and, for the first time, identify and characterize the companions to M3D and M3F as helium white dwarfs, while M3A remains undetected and M3E is shown to be a foreground contaminant. The results demonstrate the power of combining timing precision with high-resolution, multi-epoch photometry to constrain binary evolution in dense cluster environments, though limitations in photometric precision for the faint companions leave some parameters only loosely constrained. The study also highlights how dynamical cluster history and diffusion-induced hydrogen-shell flashes influence WD cooling ages, and points to future opportunities with next-generation facilities to obtain tighter mass and inclination measurements.

Abstract

The study of binary millisecond pulsars (MSPs) in globular clusters (GCs) is a key ingredient to study binary and stellar evolution under extreme conditions. In this context, an accurate analysis of the optical emission, which is mostly dominated by the companion star, is essential for a comprehensive characterization of these systems and their role within their environment. In this work, we present a multi-wavelength investigation of five binary MSPs in the Galactic GC M3 (NGC 5272) using archival Hubble Space Telescope (HST) data. Our analysis builds on the timing solutions obtained with the FAST radio Telescope by Li et al. (2024). For each MSP, we carry out precise astrometric cross-matching with the accurate radio positions to identify potential counterparts. When a match is found, we analyse its location in the colour-magnitude diagrams and compare the results with updated binary evolution models to infer the system properties. We confirm the identification of the optical companion to M3B, matching the source previously reported by Cadelano et al. (2019), and successfully identify and characterize the optical companions to M3D and M3F. All three are consistent with helium white dwarfs, as expected from the canonical formation scenario. For M3A and M3E, no reliable counterparts are found, but we place strong upper limits on the brightness and mass of the undetected companion. In the case of M3E, we detect a red object near the radio position in two F814W observations; however, astrometric measurements over a 15-year baseline reveal a significant proper motion inconsistent with cluster membership, identifying the source as a foreground contaminant. This study highlights the effectiveness of combining precise radio timing with deep, multi-band HST images to uncover and constrain the nature of MSP companions in GCs, offering insights into their formation and evolutionary histories.

Figures (12)

• Figure 1: Footprints of the FoVs covered by the HST observations listed in Table \ref{['dataset']}. The underlying stars come from the Gaia DR3 dataset Gaia2023 centerd on M3, obtained using the Gaia G-band magnitude to scale the size of the data points. Each FoV is color-coded according to the GO proposal number, listed in Table \ref{['dataset']}. The pink dots represent the position of each MSP and their name is represented. The black cross marks the center of the cluster (R.A. $=205\degree\!.548416$, Dec. $=28\degree\!.377277$) and the yellow circle is the core radius $\mathrm{r_c}=0.98$ pc. Both values are taken from the 2023 version of the GC catalog provided by Baumgardt2020.
• Figure 2: color-magnitude diagram in the ($m_{\rm F275W}$, $m_{\rm F275W} - m_{\rm F336W}$) plane. Black points represent stars from our catalog with good photometric quality. The optical counterparts to M3B, M3D, and M3F are marked with orange, yellow and green squares, respectively. Right panel: Same CMD as shown on the left, but zoomed in to focus on the regions around the counterparts. Error bars for each counterpart reflect the corresponding photometric uncertainties. Overplotted are the He WDs cooling tracks from Cadelano2020. These correspond to stellar masses, ordered from right to left, between $0.17\,\hbox{$\mathrm{M}_\odot$}$ and $0.46\,\hbox{$\mathrm{M}_\odot$}$
• Figure 3: Finding charts for the MSP M3B. The panels display _drc images in the following HST filters, from left to right: WFC3/UVIS F275W and F336W (proposal ID 12605), ACS/WFC F555W (proposal ID 10008), and F814W (proposal ID 10775). Each panel shows a $1\arcsec \times 1\arcsec$ field of view centered on the radio position of the pulsar. The magenta cross marks the nominal position of the MSP, while the magenta circle (radius $0.\!\arcsec1$) indicates the search radius. North is up, and East is to the left.
• Figure 4: Left-hand corner plots: Constraints on the mass, cooling age, and effective temperature of the companion star to M3B are presented. The 1D histograms show the marginalised likelihood distributions for each parameter, with the solid orange and dashed black lines indicating the 50th, 16th, and 84th percentiles, respectively. These percentiles represent the best-fit values and associated uncertainties. In the 2D histograms, contours correspond to the 1$\sigma$, 2$\sigma$, and 3$\sigma$ levels, with the best-fit values marked by orange points and lines. The derived values for mass, cooling age, and temperature are reported at the top of each 1D distribution panel. Right-hand corner plot: Constraints on the NS mass and the orbital inclination angle of M3B. The one-dimensional histograms represent the marginalised probability distributions for these two parameters, with the solid green and dashed black lines indicating the best-fit values and their associated uncertainties. The bottom-left panel shows the joint two-dimensional posterior probability distribution, with contours representing the 1$\sigma$, 2$\sigma$, and 3$\sigma$ confidence levels.
• Figure 5: Same as Fig. \ref{['chartM3B']} but for M3D.