2025MNRAS.541.2405B

The chemical abundance patterns of near-pristine objects provide important constraints on the properties of the first generations of stars in the Universe. We present the chemical abundances of five very metal-poor ([M/H]$< -2.5$) sub-damped Ly$\alpha$ systems (subDLAs) covering the redshift range $3.6< z< 4.3$, identified with the XQ-100 survey. We find that the subDLAs in our sample show consistent chemical abundance patterns (in particular [C/O], [Al/O], and [Fe/O]) with those of very metal-poor DLAs. Based on Voigt profile fitting, the chemical abundance ratios [C/O], [Al/O], and [Si/O] of individual velocity components in at least three of the subDLAs show some intrinsic scatter. In order to verify these chemical inhomogeneities in absorption components, we present a novel method for computing ionization corrections (ICs) on a component-by-component basis and show that ICs alone cannot explain the variations in [C/O], [Al/O], and [Si/O] between components of the same absorber at $\approx 2\sigma$ significance. Comparing the observed abundance ratios to the simulated core-collapse supernovae yields of early stellar populations, we find that all individual components of the subDLAs appears to be enriched by progenitor masses of $\lesssim 30$ M$_\odot$. The observed inhomogeneities between components can be reproduced by differences in the progenitor mass or supernova explosion energy. As such, the observed chemical inhomogeneities between components can be explained by poorly mixed gas from different nucleosynthetic events.