An 18-year Fermi-LAT stacking limit on GeV $γ$-ray emission from particle-accelerating colliding-wind binaries

Abstract

The wind-collision regions of massive colliding-wind binaries (CWBs) accelerate relativistic particles, as shown by their non-thermal radio synchrotron and, in $η$ Car, hard X-ray emission. Whether CWBs emit GeV $γ$rays as a population is unknown: only $η$ Car is unambiguously detected by the Fermi Large Area Telescope (LAT). We analyze 17.8 yr of Fermi-LAT data above 1 GeV (where the sharp point-spread function controls plane confusion) at the 61 confirmed (list A) particle-accelerating CWBs (PACWBs) of De Becker & Raucq, using a two-dimensional (photon index $\times$ flux) likelihood scan against 200 control fields matched in latitude, local source density, and diffuse intensity. Removing systems whose lines of sight coincide with bright catalogue $γ$-ray sources leaves a clean, mutually isolated sample of 6, whose stack is consistent with the control-field null ($p=0.83$): no evidence for collective GeV emission. Retaining those systems instead yields a spurious $\gtrsim3.5σ$ excess from chance catalogue coincidences, a caveat for Galactic-plane stacking. The resulting 95% limit on the mean per-source flux, $F(>1\,\mathrm{GeV})\lesssim1.1\times10^{-11}~\mathrm{ph\,cm^{-2}\,s^{-1}} (Γ=2)$, implies a $γ$-ray production efficiency $η=L_γ/L_{\rm wind}\lesssim4\times10^{-6}\,(d/\mathrm{kpc})^2$, about two orders of magnitude below $η$ Car at the sample's median distance. A representative single-zone model then bounds the electron acceleration efficiency in CWB wind-collision regions unless their magnetic field is comparable to or above the magnetic-photon equipartition value. Among the high-latitude systems accessible to a clean stack, $η$ Car thus appears to be a singular object rather than the brightest member of an emerging population.