Multiwavelength identification of millisecond pulsar candidates in the Galactic bulge

TL;DR

This work targets millisecond pulsar (MSP) candidates in the Galactic bulge by leveraging multiwavelength information to identify promising X-ray sources from Chandra data. By cross-matching with optical, UV, infrared, and radio catalogs, and incorporating a new VLA L-band mosaic, the authors prune the candidate pool from about 3158 to 1422, highlighting five strong MSP candidates with radio counterparts for follow-up pulsation searches. The approach uses a combination of the compact-object infrared criterion, low optical/UV brightness, and radio associations to filter contaminants, and discusses the limitations posed by radio sensitivity in the crowded Galactic center. The study demonstrates that multiwavelength cross-matching is an effective strategy to prioritize bulge MSP searches, while emphasizing the need for deeper radio imaging and timing observations to confirm MSP identifications and constrain the bulge MSP population relevant to the Fermi GeV excess.

Abstract

The existence of a population of millisecond pulsars in the Galactic bulge is supported, along with other evidence, by the Fermi GeV excess, an anomalous γ-ray emission detected almost 15 years ago in the direction of the Galactic center. However, radio surveys searching for pulsations have not yet revealed bulge millisecond pulsars. Identifying promising bulge millisecond pulsar candidates is key to motivating pointed radio pulsation searches. Candidates are often selected among steep-spectrum or polarized radio sources, but multiwavelength information can also be exploited: The aim of this work is to pinpoint strong candidates among the yet unidentified X-ray sources. We investigated the multiwavelength counterparts of sources detected by the Chandra X-ray observatory that have spectral properties expected for millisecond pulsars in the Galactic bulge. We considered that ultraviolet, optical, and strong infrared counterparts indicate that an X-ray source is not a bulge pulsar, while a radio or a faint infrared counterpart makes it a promising candidate. We identify a large population of more than a thousand X-ray sources without optical, ultraviolet, or strong infrared counterparts. Among them, five are seen for the first time in unpublished radio imaging data from the Very Large Array. We provide the list of promising candidates, for most of which follow-up pulsation searches are ongoing.

Figures (3)

• Figure 1: Positions of MSP candidates after successive selections: conservative selection of 2021PhRvD.104d3007B (dark blue) after removing those with optical (Section \ref{['sec:cand_opt']}, blue), UV (Section \ref{['sec:cand_uv']}, teal), strong NIR (Section \ref{['sec:cand_nir']}, green), and FIR (Section \ref{['sec:cand_fir']}, yellow) counterparts. Some of these sources have a potential VLA (radio) counterpart and/or are compact-object (CO) candidates (red crosses). Sources from the conservative selection of 2021PhRvD.104d3007B for which we found an optical counterpart therefore appear as single dark blue circles; those without an optical counterpart but with a UV one appear as a blue circle with a dark blue ring, etc. The shaded region shows the approximate coverage of the VLA mosaic (orange). The spatial distribution of the candidates is correlated with the sensitivity threshold of Chandra.
• Figure 2: Association of Chandra MSP candidates with VLA sources for our top priority candidates. The red crosses show the position of MSP candidates detected by Chandra. The red circle around them has a radius of err_ellipse_r0. The blue crosses indicate the position of radio sources detected by PyBDSF (configuration ii). When the blue cross falls inside the red circle, we consider the X-ray and radio sources to be associated. The colored background shows the VLA mosaic data in units of $\mu$Jy/beam in $\sim$20 arcsec-side squares around the X-ray sources. The color-scale is linear between 0 (purple) and the peak flux of the associated radio source quoted in Table \ref{['tab:radio_exclusion']} (yellow). From top left to bottom right, the Chandra sources are 2CXO J173946.6--282913, 2CXO J174007.6--280708, 2CXO J174011.5--283221, 2CXO J174017.3--282843, and 2CXO J174053.7--275708. Other sources are shown in Figure \ref{['fig:vla_ctp_appendix']}.
• Figure 3: Same as Figure \ref{['fig:vla_ctp']}, but for (from top left to bottom right): 2CXO J174000.6--274816, 2CXO J174810.0--285650, 2CXO J174602.4--284308, 2CXO J174616.3--284739, 2CXO J173801.2--281352, 2CXO J174000.7--274859, 2CXO J174309.3--292857, and 2CXO J174343.4--291358. These sources were not selected for follow-up studies because they are either an AGN (top left), not compact (top right), noise (second line), have an optical counterpart (third line) or have a relatively strong and previously known radio emission (bottom line). The color scale for the AGN is linear between 0 and 8522 $\mu$Jy/beam, the maximal peak flux in the box.