2024A&A...691A..21Z

Extremely metal-poor (EMPs, [Fe/H] < -3.0) and carbon-enhanced EMP (CE-EMP, [Fe/H] < -3.0 and [C/Fe]> + 1.0) stars are crucial for understanding the chemical evolution and early formation of the galaxies. Current research on EMP stars is limited by small samples, and ultraviolet (UV) band spectra are lacking. The China Space Station Telescope (CSST) will provide high-quality, low-resolution spectra across wavelengths from near-ultraviolet to near-infrared, with a limiting magnitude of about 21 mag. These will help in identifying EMPs and CE-EMP candidates in the distant Milky Way and nearby galaxies. The present study first uses the simulated CSST spectra to quantitatively evaluate the contribution of UV band spectra in predicting [Fe/H], [α/Fe], and [C/Fe]. The results indicate that UV band spectra reduce the mean absolute error by 0.04, 0.04, and 0.03, respectively, with a σ of 0.02, 0.03, and 0.04, respectively. Further spectral analysis reveals that the EMP stars show unique spectral-line features in the UV band, with significant differences compared to stars of higher metallicity, which could help in identifying EMP stars. Additionally, we tested the impact of UV band spectra on identifying EMP and CE-EMP stars at different noise levels and find that models including UV band spectra improve the identification of EMP and CE-EMP stars in terms of accuracy, recall, and F1 score. At low signal-to-noise ratio (S/N), the model including the UV band achieves a recall of 0.94, significantly higher than the model without the UV band (Recall = 0.48), doubling the capability to identify EMP stars. As the S/N increases, the inclusion of the UV band maintains high recall. This suggests that UV spectra in future large surveys could reduce the risk of missing potentially interesting stellar candidates, ensuring a more comprehensive identification of all possible EMP and CE-EMP star candidates.