W. C. Haxton, Xing Liu, Annie McCutcheon, Anupam Ray

We argue that $^{23}$Na is a potentially significant source of galactic axions. For temperatures $\gtrsim 7 \times 10^8$K -- characteristic of carbon burning in the massive progenitors of supernovae and ONeMg white dwarfs -- the 440 keV first excited state of $^{23}$Na is thermally populated, with its repeated decays pumping stellar energy into escaping axions. Odd-A nuclear abundances are typically very low in high-temperature stellar environments (or absent entirely due to burn-up). $^{23}$Na is an exception: $\approx 0.1 M_\odot$ of the isotope is synthesized during carbon burning then maintained at $\approx 10^9$K for times ranging up to $6 \times 10^4$y. Using MESA simulations, a galactic model, and sampling over progenitor masses, locations, and evolutionary stages, we find a continuous flux at earth of $\langle \phi_a \rangle \approx 22$/cm$^2$s for $g^\mathrm{eff}_{aNN} = 10^{-9}$. Some fraction of these axions convert to photons as they propagate through the galactic magnetic field, producing a distinctive 440 keV line $\gamma$ ray detectable by all-sky detectors like the Compton Spectrometer and Imager (COSI). Assuming a 1$\mu$G galactic magnetic field and a sufficiently light axion mass, we find that COSI will be able to probe $| g_{aNN}^\mathrm{eff} g_{a \gamma \gamma} | \gtrsim1.8 \times 10^{-22}$ GeV$^{-1}$ at $3\sigma$ after two years of surveying.