Long-term timing of the relativistic binary PSR J1906+0746

TL;DR

PSR J1906+0746 is a young relativistic binary whose long-term timing is used to test GR in DNS-like systems. We combine 18+ years of multi-telescope ToA data and model red timing noise with a Fourier-basis Gaussian process, achieving phase-coherent timing across the full span. The study yields precise post-Keplerian parameters: $\dot{\omega}=7.5841(2)$ deg/yr, $\gamma=4.59(2)\times 10^{-4}$ s, and $\dot{P}_{\rm b}=-5.65(2)\times 10^{-13}$ s s$^{-1}$, implying $M_{\rm total}=2.6133(1)\,M_\odot$ and component masses $M_{\rm p}=1.316(5)\,M_\odot$, $M_{\rm c}=1.297(5)\,M_\odot$ under GR. A new measurement of $\dot{x}=-1.8(6)\times 10^{-13}$ lt-s s$^{-1}$ hints at spin-orbit coupling; if confirmed, it could indicate a massive fast-rotating white dwarf companion, though a neutron-star companion remains possible. The timing also reveals a large glitch near MJD 56664 and a quasi-periodic residuals feature near 2 years, with future timing needed to confirm the companion nature and assess gravity tests.

Abstract

We conducted a timing analysis of over 18 years of data on the young ($τ_{\rm c} = 112$ kyr, $P = 114$\,ms) relativistic binary PSR~J1906+0746, using six radio telescopes: Arecibo, FAST, Green Bank, Lovell, MeerKAT, and Nançay. This pulsar is known to orbit a compact high-mass companion with a period of 3.98\,hrs in a mildly eccentric orbit ($e = 0.085$). By combining all data and maintaining a coherent timing solution over the full span, we obtained a more precise measurement of the advance of periastron, $\dotω = 7.5841(2)$\,$°$\,yr$^{-1}$, the Einstein delay, $γ= 4.59(2) \times 10^{-4}$\,s, and the secular change in orbital period, $\dot{P}_{\rm b} = -5.65(2) \times 10^{-13}$\,s\,s$^{-1}$. Assuming the validity of general relativity, we obtain a total mass of 2.6133(1)\,$M_{\odot}$ and component masses of 1.316(5) $M_{\odot}$ for the pulsar and 1.297(5) $M_{\odot}$ for the companion, consistent with a double neutron star system. However, when fitting for the secular change in the projected semi-major axis we obtain $\dot{x} = -1.8(6) \times 10^{-13}$\,s\,s$^{-1}$, the component masses are shifted by $\sim 3.5σ$, which is expected from the correlation of $\dot{x}$ and $γ$. The $\dot{x}$ has a similar magnitude to that observed in PSR~J1141$-$6545, which is due to spin-orbit coupling; if confirmed, it would indicate that, as in the latter system, the companion of the PSR~J1906+0746 system is a massive fast-rotating white dwarf formed before the pulsar. Additionally, we report and characterize a large glitch near MJD 56664, with a fractional frequency increase comparable to those observed in the Vela pulsar.

Figures (7)

• Figure 1: Integrated pulse profiles of PSR J1906+0746 illustrating all the different profile variations seen in the DFB Jodrell (JBO), Arecibo PUPPI (AO), MeerKAT (MKT), and FAST data. The main pulse (left) and interpulse (right) are shown. For each epoch, a baseline level has been removed and the profile has been scaled such that its maximum equals unity. The dates of the observations used to obtain the profiles are indicated on the right.
• Figure 2: Phase-connected timing residuals as a function of MJD. The top panel shows the residuals we obtained as a result of the fitting procedure from RUN_ENTERPRISE obtaining phase connection throughout our whole data set without removing the timing noise. The black line denotes the timing noise model fitted, while the purple dashed line denotes the glitch epoch. The bottom panel shows the result after including up to the 300th harmonic for whitening.
• Figure 3: Power spectral density against frequency is shown by the solid blue line. The red line represents the fitted power law (Equation \ref{['eq:powerlaw']}). A peak at a frequency of 0.493 yr$^{-1}$ indicated by the vertical dash-dotted blue line, suggests evidence for a periodicity in the residuals of $\sim$2 yr.
• Figure 4: Timing residuals, $\Delta \nu$, and $\dot{\nu}$ as a function of time for PSR J1906+0746 after removing the red noise including up to the 60th harmonic for whitening. The vertical purple dashed line denotes the glitch epoch.
• Figure 5: Top panel: evolution of $\Delta \dot{\nu}$ sampled at the Nyquist frequency. Bottom panel: Power spectral density of $\dot{\nu}$.