Probing the millisecond pulsar origin of the $γ$-ray excess in the Galactic centre with LISA

TL;DR

This work assesses whether LISA can test the millisecond pulsar (MSP) origin of the Galactic center gamma-ray excess by forecasting the population of MSP–white dwarf binaries in the Galactic bulge detectable in the millihertz gravitational-wave band. Using two formation channels—accreted MSPs from disrupted globular clusters and in situ binaries from isolated binary evolution—the authors simulate bulge MSP populations and compute LISA detectability, including the ability to measure chirp masses for the most massive or high-frequency systems. They find that only a small fraction of the underlying MSP population would be detectable by LISA ($f^{\rm LISA}\approx 10^{-5}$–$10^{-4}$), with accreted binaries peaking at $\mathcal{M}\sim 0.4 M_\odot$ and in situ binaries at $\mathcal{M}\sim 0.7$–$1.1 M_\odot$, though these contrasts depend on modelling assumptions. A major challenge is distinguishing MSP binaries from the dominant WDWD background using GW data alone; the study highlights a practical path forward: follow-up radio observations with the Square Kilometre Array (SKA) for target confirmation, combined with multi-messenger data from CTA and future GW observatories, to test the MSP-origin scenario for the gamma-ray excess.

Abstract

The gigaelectronvolt $γ$-ray excess observed towards the Galactic centre remains unexplained. While dark matter annihilation has long been considered a leading explanation, an alternative scenario involving a large population of millisecond pulsars remains viable. Testing this hypothesis with electromagnetic observations is difficult, as pulsar searches in the bulge are strongly affected by interstellar scattering, high sky temperature, and source confusion. We investigate whether gravitational-wave observations with the Laser Interferometer Space Antenna (LISA) could provide an independent probe of the millisecond pulsar binary population in the Galactic bulge in the future. We constructed synthetic populations of detached millisecond pulsar-white dwarf binaries under two illustrative formation scenarios: an accreted scenario, in which systems are deposited by disrupted globular clusters, and an in situ scenario, in which binaries form through isolated binary evolution. In both cases, only $10^{-5}$-$10^{-4}$ of the underlying bulge population is detectable by LISA. Still, even a few detections would imply tens to hundreds of thousands of unseen systems. Accreted binaries are expected to have lower chirp masses ($\sim$0.4 M$_\odot$), while in situ binaries produce more massive companions ($\sim$0.9 M$_\odot$), though part of this contrast reflects our modelling assumptions. LISA will measure binary frequencies with high precision, but chirp masses can only be determined for the most massive or highest-frequency systems. Thus, identifying millisecond-pulsar binaries among the far more numerous double white dwarfs will be challenging, as their gravitational-wave signals alone are indistinguishable. However, coordinated follow-up with the Square Kilometre Array of LISA-selected targets could directly test the millisecond-pulsar explanation of the $γ$-ray excess.

Figures (4)

• Figure 1: Orbital periods and companion masses for MSPs compiled from the ATNF Pulsar Catalogue v2.6.0. The companion mass shown is the median mass, computed assuming an orbital inclination of $i = 60^\circ$. All data points (blue cross symbols) correspond to radio-detected MSPs. Among them, $\gamma$-ray sources are highlighted with green downward-pointing triangles, and those located in globular clusters (GCs) are marked with orange upward-pointing triangles. Grey error bars indicate lower and upper limits on WD companion masses. For reference, the theoretical orbital period–companion mass relation from TaurisSavonije1999AA for Population I stars is shown as a solid red line. The minimum gravitational-wave frequency detectable by LISA is shown as a dashed black line.
• Figure 2: Top panel: Distribution of companion masses for the MSP sample. The solid red lines show a power-law distribution with a slope of $-4$, plotted for comparison. Bottom panel: Distribution of binary orbital periods for the MSP sample. Overplotted are the estimated pulsar detection efficiencies in radio surveys, based on the modelling by Bagchi2013MNRAS. Two search strategies are shown: $\gamma_1$ for a standard (non-accelerated) search, and $\gamma_2$ for an acceleration search. These detection efficiency curves are computed assuming the following orbital parameters: eccentricity $e = 0$, $M_\mathrm{MSP} = 1.4$ M$_\odot$, integration time $T\mathrm{obs} = 1000$ s, WD companion mass derived from the TaurisSavonije1999AA relation, spin period of 3 ms, orbital inclination $i = 60^\circ$, and longitude of periastron $\Omega = 30^\circ$.
• Figure 3: Detectability of a binary system containing a 1.4 M$_\odot$ MSP with LISA, shown as a function of companion mass and gravitational-wave frequency (bottom x-axis) or orbital period (top x-axis). Grey contours indicate LISA’s 4-year signal-to-noise threshold ($\rho_{\rm 4\,yr} = 7$) for sources located at 5 (dark grey), 8 (grey), and 11 kpc (light grey). Dashed horizontal lines mark approximate boundaries between different classes of MSP binaries: black widows (low-mass companions, $M \lesssim 0.06$ M$_\odot$), redbacks and He WDs (intermediate-mass companions, $0.06 \lesssim M \lesssim 0.5$ M$_\odot$), and more massive donors such as CO/ONeMg WDs or He stars ($M \gtrsim 0.5$ M$_\odot$). The grey solid curve marks the Roche-lobe overflow (RLOF) boundary for WD donors, indicating the onset of stable mass transfer. Observed MSP binaries are shown as coloured symbols, following the same classification as in Fig. \ref{['fig:orbital_periods']}. UCXBs lying near the WD RLOF line are shown as light blue pentagons. The two-segment grey arrow illustrates the expected evolutionary track of close binaries.
• Figure 4: Detectable MSPWD binaries in the chirp mass–gravitational-wave frequency plane for both formation scenarios considered in this study. The in situ population is shown with orange stars, while systems from the accreted scenario are marked with green downward-pointing triangles. For comparison, the underlying population of detectable WDWD binaries is shown in blue-grey. Contours indicate regions of constant relative uncertainty in the chirp mass ($\sigma_{\mathcal{M}}/\mathcal{M}$) at levels of 1, 0.1, and 0.01, highlighting LISA’s ability to constrain binary parameters. The hatched region marks the part of parameter space where the chirp mass cannot be reliably measured ($\sigma_{\mathcal{M}}/\mathcal{M} > 1$). The top axis shows the corresponding orbital period.