Fe Abundances of Early Galaxies at z=9-12 Derived with Deep JWST Spectra

We derive Fe-abundance ratios of 7 galaxies at $z=9-12$ with $-22<M_{\mathrm{UV}}<-19$ whose JWST/NIRSpec spectra achieve very high signal-to-noise ratios, $\mathrm{SNR}=60-320$, at the rest-frame UV wavelength. We fit stellar population synthesis model spectra to these JWST spectra, masking out nebular emission lines, and obtain Fe-abundance ratios of $\mathrm{[Fe/H]}=-1-0$ dex for 5 galaxies and upper limits of $\mathrm{[Fe/H]}\sim-2-0$ dex for 2 galaxies. We compare these [Fe/H] values with the oxygen abundances of these galaxies ($7.4<12+\log{\mathrm{(O/H)}}<8.4$) in the same manner as previous studies of $z\sim2-6$ galaxies, and derive oxygen-to-iron abundance ratios [O/Fe]. We find that 2 out of 7 galaxies, GS-z11-0 and GN-z11, show Fe enhancements relative to O ($\mathrm{[O/Fe]}<0$ dex), especially GS-z11-0 ($z=11.12$) with a Fe enhancement ($\mathrm{[O/Fe]}=-0.68_{-0.55}^{+0.37}$ dex) beyond the solar-abundance ratio at $\sim2\sigma$. Because, unlike GS-z11-0, GN-z11 ($z=10.60$) may be an AGN, we constrain [O/Fe] via FeII emission under the assumption of AGN and confirm that the Fe enhancement is consistent even in the case of AGN. While [O/Fe] values of most galaxies are comparable to those of core-collapse supernovae (CCSNe) yields, the Fe enhancements of GS-z11-0 and GN-z11 are puzzling. We develop chemical evolution models, and find that the Fe enhancements in GS-z11-0 and GN-z11 can be explained by 1) pair-instability supernovae/bright hypernovae with little contribution of CCSNe or 2) Type-Ia supernovae with short delay time ($\sim30-50$ Myr) with a top-light initial mass function.