Signature of a massive rotating metal-poor star imprinted in the Phoenix stellar stream

The Phoenix stellar stream has a low intrinsic dispersion in velocity and metallicity that implies the progenitor was probably a low mass globular cluster. In this work we use Magellan/MIKE high-dispersion spectroscopy of eight Phoenix stream red giants to confirm this scenario. In particular, we find negligible intrinsic scatter in metallicity ($\sigma(\mathrm{[Fe~II/H]}) = 0.04^{+0.11}_{-0.03}$) and a large peak-to-peak range in [Na/Fe] and [Al/Fe] abundance ratios, consistent with the light element abundance patterns seen in the most metal-poor globular clusters. However, unlike any other globular cluster, we also find an intrinsic spread in [Sr II/Fe] spanning $\sim$1 dex, while [Ba II/Fe] shows nearly no intrinsic spread ($\sigma(\mathrm{[Ba~II/H]}) = {0.03}^{+0.10}_{-0.02}$). This abundance signature is best interpreted as slow neutron capture element production from a massive fast-rotating metal-poor star ($15-20 \mathrm{M}_\odot$, $v_\mathrm{ini}/v_\mathrm{crit} = 0.4$, $[\mathrm{Fe/H}] = -3.8$). The low inferred cluster mass suggests the system would have been unable to retain supernovae ejecta, implying that any massive fast-rotating metal-poor star that enriched the interstellar medium must have formed and evolved before the globular cluster formed. Neutron capture element production from asymptotic giant branch stars or magneto-rotational instabilities in core-collapse supernovae provide poor fits to the observations. We also report one Phoenix stream star to be a lithium-rich giant ($A(\mathrm{Li}) = 3.1 \pm 0.1$). At $[\mathrm{Fe/H}] = -2.93$ it is among the most metal-poor lithium-rich giants known.
Comment: Accepted to ApJ 2021-07-09