Two Possible Optical--X-Ray Anti-Correlations of PSR J1023+0038

TL;DR

This work analyzes simultaneous X-ray and optical (B-band) data from PSR J1023+0038, a transitional redback MSP, to test irradiation mechanisms by searching for correlations between the optical modulation amplitude $MA$ and high-energy emission, complemented by γ-ray analysis from Fermi-LAT. It finds a general $MA$–X-ray anti-correlation whose significance depends on the analysis method, and a stronger anti-correlation when using X-ray medians; γ-ray shows no robust relation to $MA$. A secondary result reveals a low-mode anti-correlation between EPIC X-ray low states and OM optical flux in several observations, suggesting episodic or geometry-driven irradiation changes. The findings imply that X-ray emission is not the primary source heating the companion; instead, pulsar winds and disk-inner-flow–driven shocks likely modulate irradiation, with variability possibly linked to outflows or disk precession. Deeper, coordinated multi-wavelength campaigns are required to fully understand the heating and mode-switching in this system.

Abstract

X-ray emission is generally believed to be one of the major heating sources for the optical modulation in redback pulsar binaries as we have seen similar phenomena in many low mass X-ray binaries (LMXBs). While, e.g., MeV/GeV gamma-rays from the neutron stars are also possible heating sources, X-ray observations are currently much more sensitive, and therefore, joint optical--X-ray data are observationally unique on the irradiation mechanism investigation. Using 18 X-ray/B-band simultaneous XMM-Newton observations (717 ks in total) of the redback system PSR J1023+0038 taken during the LMXB state, we find a general trend that the amplitude of the B-band orbital modulation was lower when the observed X-ray flux was higher. Depending on the analysis method adopted, the statistical significance of the anti-correlation can be from 1.7sigma to 3.1sigma. We also extended the analysis to the GeV gamma-ray band using the Fermi-LAT data, but the result is insignificant to claim any relations. Moreover, another X-ray/optical correlation regarding the low modes of the system was found in some of the \textit{XMM-Newton} observations, and the astrophysical reason behind is currently unclear yet. These anti-correlations likely suggest that the irradiation is generally stronger when the X-ray flux is in a fainter state, indicating that there is a more dominant irradiation source than the X-ray emission.

Figures (17)

• Figure 1: The B-band light curves of Obs. C and J. The translucent dots and the cross/dashed lines refer to the 10s-binned and 400s-binned OM data, respectively (red/top for Obs. C and blue/bottom for Obs. J). The orange and magenta solid curves are the best-fit models of Obs. C and J, respectively. All the data relevant to Obs. C are shifted upwards by 10 ct/s to clarify the two data sets.
• Figure 2: The X-ray flux bimodal distribution of J1023 (Obs. A) with the best-fit double Gaussian model.
• Figure 3: The EPIC (red) and detrended OM (blue) light curves of J1023 that show the non-detection (upper panel; part of Obs. B) and detection (lower panel; part of Obs. D) of the low-mode anti-correlation. The OM light curves were re-binned with a 20s bin size for better visualization. The gray shadows indicate the identified X-ray low mode of J1023.
• Figure 4: The non-low-mode occurrence rate vs. optical MA. The solid line shows a possible anti-correlation between the non-low-mode occurrence rate and the optical MA.
• Figure 5: The X-ray mean count rate (top: no filtration, middle: filtered by condition (1), bottom: filtered by condition (2)) vs. optical MA plot. The mean error bars are not visible because the errors are too small (Table \ref{['tab:epicom_exp']}). The solid lines are used to visualize the anti-correlation by fitting a linear function, and the transparent red crosses refer to the observations that are filtered by the corresponding conditions. The anti-correlation significances by using mean count rate are between 1.7--2.6$\sigma$ (see Sections \ref{['sec:X']}).